Continuous Dynamic Optimization: Impacts on ADA Paratransit Services (2023)

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.

65   Considerations About Whether or Not to Obtain/Deploy CDO Technology: Pros and Cons The considerations described in this section are synthesized from the survey responses, case examples, and interviews. Benefits of CDO Technology Virtually all of the transit agencies involved in both the survey and the case examples were enthusi- astic about the use of CDO as a way to automate proactive dispatching and in some cases, the sched- uling function as well. Common motivations for adopting the technology included the following: • To improve productivity. Proactive dispatching is one of the few aspects of CDO that positively impacted both productivity and on-time performance. Eight of the transit agencies interviewed specifically acquired technology that had CDO capability to automate this process. In doing so, these agencies better addressed—and counteracted—decreases in productivity that typically arise from driver callouts and downed vehicles, late cancellations and no-shows, incidents and breakdowns, and vehicles running late because of unforeseen traffic congestion and other fac- tors. Six of the 11 agencies reported that CDO did in fact improve productivity. The average improvement in productivity was 17 percent, and one transit agency achieved a 31 percent improvement in productivity. • To reduce overall and per-trip costs. Part and parcel with improving productivity is serving more trips with the same funding, serving the same number of trips for less, or both. In addition to increasing the number of shared rides, several of the transit agencies use CDO to reduce the number of runs and revenue service hours. Six of the agencies (those that provided before-and- after data) indicated that the change in operating costs ranged from a 30 percent decrease to a 4 percent increase, with the average change in operating costs amounting to a 13 percent decrease. The reduction in operating costs was largely the result of a decrease in the number of vehicles, revenue vehicle hours needed, or both, while the reduction in unit operating costs stemmed from being able to serve more trips with fewer resources. • To improve on-time performance. Seven of the 11 the transit agencies indicated that improving on-time performance was another motivator in acquiring technology with CDO. Six of the agencies indicated that their on-time performance did, in fact, improve, with one transit agency indicating a 9 percent improvement. For agencies that experienced a decrease in on-time perfor- mance (with an average decrease of 1 percent), the underlying reason was the significant increase in productivity. That is, these agencies indicated that the very slight decrease in on-time perfor- mance was worth the significant increase in productivity and the resulting decrease in operating cost. This is the kind of decision that every transit agency faces with its ADA paratransit service— the balance between service/cost efficiency and service quality. And in the case of the balance C H A P T E R   4 Conclusions and Lessons Learned

66 Continuous Dynamic Optimization: Impacts on ADA Paratransit Services tipping in favor of productivity, the motivation was to be able to serve more trips in a cost- efficient manner without really compromising the quality of the service. • To address dispatching problems. Eight of the 11 transit agencies saw CDO as a way to improve the quality of the dispatching. Four of the 11 agencies indicated that prior to the instal- lation of CDO, their dispatch staff was overburdened, while one agency indicated that its pro- active dispatcher did not have the requisite skill set to be successful at that task. The idea behind proactive dispatching is to identify and solve problems before they manifest themselves in real time. The largest of the agencies, CapMetro, indicated that it automated service-day optimiza- tion simply because of the size of its service (2,000 trips per day). Even a proactive dispatcher looking for problems ahead could not get around to solving all the issues in advance, and the real-time problems accumulated. CapMetro indicated that 20 percent of its trips change in the course of the service day as a result of the factors mentioned previously, and on some days, as much as 40 percent of the trips change. Simply put, an automated approach to service-day opti- mization through CDO can do better job at proactive dispatching than a person can. More- over, eight (73 percent) of the transit agencies reported a reduction in real-time issues. With this reduction (and CDO handling the proactive dispatching), dispatchers could better address the real-time problems because they were fewer and because more dispatchers were available to help out. In addition, a few of the transit agencies indicated that the objectivity of the opti- mization addressed some long-standing dispatcher-driver favoritism issues that plague some paratransit services. While not motivators for the initial purchase of the technology, the following other benefits have accrued from CDO: • A reduction in, or repurposing of, scheduling/dispatch staff and a reduction in scheduler/ dispatcher labor costs. Only three transit agencies were able to reduce the number of dispatchers, two transit agencies were able to reduce their scheduling staff, and three agencies reported a reduction in scheduling/dispatch labor costs resulting from use of CDO. The follow-up inter- views indicate that the reason more agencies were unable to reduce the number of dispatchers had to do with the need for dispatch coverage, especially with the smaller transit agencies. Still, several of the transit agencies that were not able to reduce their dispatch staff indicated that they were able to expand the duties of the staff, adjust their focus (to processing no-shows and solving problems—those not solved by CDO—in real time), or both. For MET Transit in Billings, MT, CDO enabled the hiring of a fourth dispatcher for Saturday dispatching of both the fixed- route and paratransit services, as well as better coverage for when a dispatcher is sick or goes on vacation. Gold Coast Transit District in Oxnard/West Ventura County, CA, reported that CDO allowed it to grow new service offerings (e.g., meal delivery, an expanded service area, and late-night service) of up to 20 percent in some months without the contractor having to provide additional dispatchers. • Reduction in customer complaints. Across the four transit agencies that provided before-and- after data, the change in customer complaints as a result of CDO ranged from a 100 percent decrease to a 50 percent increase. The average change in customer complaints was a 56 percent reduction. An additional agency indicated that its complaint level went from “many” to “few.” The reduction in complaints correlated with the transit agencies whose on-time performance improved, while the increase correlated with the agencies whose on-time performance declined (because of the significant increase in shared rides). Applicability of CDO The service model did not seem to have a significant impact on the applicability of CDO, as shown in the following examples: • Service eligibility and commingling. CDO worked as well in the transit agencies that used it to support ADA paratransit service alone as it did in agencies that used it to support coordinated/

Conclusions and Lessons Learned 67   consolidated services that commingled ADA paratransit with trips sponsored by human service agencies (including Medicaid NEMT trips) and general-public trips. For example, Gold Coast Transit District in Oxnard/Western Ventura County, CA, commingles senior trips with ADA paratransit trips; PSTA in St. Petersburg, FL, commingles transportation-disadvantaged trips with ADA paratransit trips; and Citibus in Lubbock, TX, commingles general-public trips and Medicaid NEMT trips with ADA paratransit trips. In Lubbock, TX, and Summit County, UT, CDO is being used for services in which the ADA service is consolidated with microtransit trips, and ADA para- transit customers take advantage of the on-demand feature of the microtransit service. • In-house versus contracted service. Based on the survey and the profiles, CDO is used both by transit agencies that directly operate service and transit agencies that retain turnkey contrac- tors and operations contractors. For example, the transit agencies in Abilene, TX; Austin, TX; Billings, MT; Lubbock, TX; Sheboygan, WI; and Wenatchee, WA, perform all call and control center functions in house. Four of these transit agencies operate the service in its entirety. The exceptions include CityLink in Abilene and Link Transit in Wenatchee, which operate most of the service in house but also use contractors. CityLink uses a local taxi company as an overflow provider (and formerly used a human service agency as a contractor as well), while Link Transit uses both a hospital to provide service in a remote area on an on-call basis and NEMT carriers as overflow providers. The transit agencies in four locations (Bowling Green, KY; Green Bay, WI; Oxnard, CA; and Summit County, UT) vest most, if not all, of the day-to-day functions with a turnkey contrac- tor. Under a turnkey contract, it is more difficult to see such daily actions unless the oversight is attentive because most turnkey contracts are paid wholly or partly on a variable cost per hour. There is no built-in incentive to reduce hours. Many contracts, however, have productivity- based bonuses, which can incentivize contractors to cut runs short if work rules allow. Service models that involve centralized scheduling and dispatching functions (with multiple operations contractors) are also a good fit for CDO because the transit agency (or a call center manager or broker) controls the dispatching function. Last, PSTA in St. Petersburg, FL, has a split structure in which it handles the rider interface (bookings and ETA calls), while First Transit handles operations for the accessible dedicated fleet with the related dispatching function, and a taxi company and UZURV provide service to ambulatory riders. This is an unusual service model because the initial assignments to the three carriers is a decision first made by the booking agent. CDO comes into play on the days leading up to the day before the trip date. Trips can be booked up to a month in advance. During this period, CDO can move trips between First Transit and the taxi company. On the service day, however, CDO excludes the taxi company from consideration. UZURV is, in effect, the overflow service when the taxi company is at capacity, but the process of assigning trips to UZURV is manual and is not included in the CDO process. • Service mix and brokering. Service mix is the combination of dedicated and non-dedicated service. The three transit agencies that use non-dedicated overflow providers are in Abilene, TX; St. Petersburg, FL; and Wenatchee, WA. In Abilene, the taxi drivers have the same tablets as the in-house drivers, so the CDO processes can consider them as well on the day of service. All of the CDO technologies used by the transit agencies can handle brokering to non-dedicated service providers that supplement the dedicated fleet or fleets. Generally, there are two ways that these technologies decide which trips to broker. The first is based on optimizing the dedi- cated fleet based on the parameters used in the scheduling algorithms. The concept here is to use a resource that a transit agency has, in effect, already paid for. The second way is based on cost to the transit agency, where cost is based exclusively on the attributes of the trip and the provider, not on how the trip fits into a dedicated run. In this approach, the technology provides inputs for the contractual cost structure of each non-dedicated service provider. This arrangement generally results in a list of trips that is sent to or brokered with the non- dedicated provider for dispatching on the day of service unless the overflow drivers are linked to the CDO system, in which case a trip can be sent directly to the driver.

68 Continuous Dynamic Optimization: Impacts on ADA Paratransit Services Indeed, CDO automates the assignment or reassignment of trips to overflow providers as a way to handle a reemerging no-show or a will-call return, or just to get a run back on schedule. At the same time, CDO can also take back a trip that had been assigned to an overflow provider should a hole appear on a dedicated run because of a cancellation (if this option is permitted by the subcontract provisions). Impact of CDO on Scheduling Practices and Policies CDO’s impact on traditional scheduling practices and advance reservation policies is one of the more interesting findings of the synthesis. • Scheduling practices. In the traditional paratransit scheduling/dispatching systems of the 1980s and the 1990s, the booking, scheduling, and dispatching functions were somewhat separate with the exception of real-time scheduling as enabled by the prevailing software. Still, even with real- time scheduling, it was commonplace for scheduling to be a separate function often performed by separate staff (except at smaller agencies) using the software to adjust the schedule one trip at a time or through batch scheduling. With the advent of CDO in late 2000s, transit agencies began to see not only the possibility of replacing much of the service-day dispatch effort with CDO but also the possibility of replacing the traditional scheduling process (and separate staff) with CDO on the day (or days) leading up to the day of service. Basically, what has happened, as evidenced by several of the transit agency profiles, is that both of these functions have been predominantly replaced with CDO, and while a transit agency can still use the more traditional paratransit scheduling and dispatch- ing systems along with CDO, the newer technologies really do not provide a choice. That is, in acquiring these technologies to support its ADA paratransit service, a transit agency would be committing to automating both the scheduling and dispatching processes. In effect, these agen- cies are committing to CDO. The use of CDO supports transit agencies that, by policy, do not provide a confirmed pickup window at booking. The CDO process develops the final, scheduled time on the evening before the service, at which time a confirmed pickup window can be communicated to the rider. If, by policy, a confirmed pickup window, based on the requested pickup time or drop-off time, is provided to the rider immediately after the trip is booked, then careful controls need to be added to the algorithms’ parameters to limit how far the pickup time can be bumped from the original requested time. • Advance reservation policies. The changing practices in scheduling and dispatching with CDO in turn enable a transit agency to have, if desired, a next-day (only) reservation policy as exem- plified by Green Bay Metro in Wisconsin. Alternatively, a transit agency can reduce the span of the advance reservations to only a few days; for example, CapMetro riders in Austin, TX, can book a trip over the phone up to three days in advance, and Link Transit riders in Wenatchee, WI, can book trips up to five days in advance. Narrowing the advance reservation period to next day only or to just a few days is a proven way to cut down on the cancellations and no-shows associated with placeholder reservations booked well in advance, especially if the optimization processes do not get underway until the day before the trip. Related Issues in Tailoring, Deploying, and Using CDO Understanding How the Parameters Work in Relation to Transit Agency Goals As one general manager phrased it, “It is very important to understand the basics of your CDO algorithm so that you know which elements of the service you can affect, which elements you can prioritize, and which parameters you can change.”

Conclusions and Lessons Learned 69   The experiences of the 11 transit agencies suggest that transit agencies in general need to understand not only how reliance on CDO may change the way the scheduling and dispatching function is carried out but also how the parameters can be shaped to provide the desired balance between service/cost efficiency and service quality. For example, in Billings, MT, MET Transit’s goal was to optimize its vehicles and drivers by reducing the number of revenue vehicle hours required and spreading trips evenly across resources while not compromising its standards for on-time performance. Service/cost efficiency is typically measured by the percentage of shared rides, the ratio of actual passenger miles to direct passenger miles, trips per revenue vehicle hour, and operating cost per trip. Service quality is typically measured by on-time performance, average travel time, violations of maximum onboard time, and the complaint frequency ratio (often measured in terms of the number of complaints per 1,000 passenger trips). Prior to tuning parameters to achieve the balance between service/cost efficiency and service quality, transit agencies must first input the hard numbers reflecting policy, such as the span of the pickup window, the default loading and unloading times for ambulatory versus non-ambulatory rides, and the maximum onboard time. These parameters are worth revisiting while transitioning to new technologies in consideration of the desired balance; it is also worth recognizing that it is possible to instruct the CDO process to relax these policies to a certain extent if this makes sense, again in consideration of the desired balance. Parameter Tuning In tuning other parameters, transit agencies are encouraged to define how important they are. For example, how important is it that the scheduled pickup time be in the confirmed pickup window, or that there be no travel times longer than the maximum onboard time? Is there any leeway there, or is this a hard and fast constraint? Other parameters linked to service/cost efficiency include minimizing total and deadhead miles traveled; minimizing vehicle runs, revenue service hours, or both; and maximizing shared rides. A parameter that is linked to both productivity and service quality is the extent to which a vehicle backtracks or zigzags to pick up or drop off another passenger (i.e., before the CDO technology assigns that potentially ride-shareable trip to another run.) With the tuning of these optimization parameters, there is no right or wrong; there are only gradations. Most commonly, the optimal setting is achieved by trial and error in a test area—that is, by re-valuing the importance of a parameter or parameters, running an optimization process, and reviewing the results. All of the systems used by the 11 transit agencies allow either their staff or the vendor’s technician to act as their proxy to identify how important each parameter is. For example, if a transit agency is consciously striving for service efficiency, it might stress the impor- tance of the service/cost parameters identified previously. If the transit agency leans more heavily toward achieving high on-time performance, these parameters would be devalued. Software systems that provide agencies with direct access to the parameters are therefore helpful in this iterative process as long as the transit agency staff is willing to put in the time and effort with proper controls. This can be done with helpful guidance from the vendor’s technicians. For the other newer technologies, in which a vendor technician has the control, the achievement of the desired balance between service/cost efficiency and service quality is subject to the technician’s interpretation of the transit agency’s direct input and his or her availability and time. That said, most of the 11 transit agencies have been happy with the vendor’s responsiveness and tuning suc- cesses. One of the agencies that will soon be transitioning to a newer technology stated that it was somewhat apprehensive about losing direct control. Green Bay Metro in Wisconsin reported that during the transition phase and after implemen- tation, its operations/technology contractor helped its staff to understand how each parameter

70 Continuous Dynamic Optimization: Impacts on ADA Paratransit Services of the technology worked to improve both productivity and on-time performance, and then made sure that the ADA paratransit service’s policies and procedures (e.g., pickup window span and maximum onboard time) were properly reflected in the technology. Almost all of the transit agencies agreed that ongoing attention and related tweaking of the parameters to address suboptimal performance are necessary. Tailoring Optimization and Subroutine Processes Beyond the parameters, transit agencies agreed that their understanding of how CDO pro- cesses can be tailored is also important. For example, the majority of the transit agencies surveyed indicated that their CDO processes involved being able to screen out or include certain trips in a particular optimization process, anchor the pickup times for selected trips, and order subroutine processes such as optimizing wheelchair trips first. Many of the agencies also indicated that their CDO processes involved the relaxation of some service quality standards. Different Categories of Optimization With some CDO technologies, there may be two categories of optimization. The first category of optimization no longer considers a vehicle run once it begins service. That is, the trips already on the run are frozen out of the optimization process, and the CDO will not schedule a new trip on that run. The second category of optimization considers all runs in service and all trips in the optimization process. Another important consideration for a transit agency is whether the optimization algorithms reflect one set of algorithms or whether the agency can identify separate processes—with a different set of parameters—for separate jobs. One vendor’s technology, for example, can provide for a limitless number of optimization routines that can be automated to run at different times and do different tasks, while the other vendors have taken a one-size-fits-all approach. Harkening back to the previous discussion about scheduling practices and advance reservation policies, a system that has a next-day-only booking policy may be less concerned about the need for different optimiza- tion routines, whereas there may be value in different routines in systems in which long lead times for reservations are permitted. Planning, Deployment, and Training Another interesting dynamic in the planning/implementation phase is whether or not certain service policies should be changed as a direct or indirect result of using CDO. For instance, one transit agency discussed the implications of changing the span and the particulars of the pickup window. This policy could be easily tested. Another agency, after setting 90 minutes as a maximum onboard travel time, adjusted this parameter to 60 minutes after a few weeks of using CDO. High Valley Transit in Summit County, UT, in implementing a consolidated ADA paratransit service under the same technology platform using CDO, opted to first keep the two services sepa- rate for six months so that each set of riders could get used to their respective service before the two services were consolidated. MET Transit in Billings, MT, reported that as with any new software implementation, there was a learning curve for both operators and dispatchers. Many of the other agencies agreed with this, some adding that it is imperative that drivers are properly trained. For example, in a procurement, one of the concepts stressed was to require that each proposer or finalist demonstrate its training plan for drivers. Some of the technologies have simulation software for training purposes that allows the driver to get used to how the tablets work, how to perform events, how to submit a no-show

Conclusions and Lessons Learned 71   request, how to retrieve navigation assistance, and so forth. One technology vendor advised not shifting all drivers to the new system at once (e.g., using tablets) so that the system and the training can be adjusted as needed before systemwide use if indeed this is feasible technologically. Lessons Learned Do Not Be Afraid of CDO, Especially Because the Benefits Can Be Substantial The Citibus general manager in Lubbock, TX, advised, “Don’t be scared to try CDO. There’s nothing really to watch out for. Embrace the opportunity to do things that are different and non- traditional. Push the envelope for the betterment of the service.” The MET Transit general manager in Billings, MT, highly recommended that other transit agencies try CDO. She said that CDO “has directly contributed to better use of dispatcher time, increased productivity, and availability of more data regarding MET Transit operations,” adding that the technology was “a significant investment, but on an annual basis, MET Transit is realizing significant cost savings.” The Green Bay Metro general manager added that CDO has been beneficial for both the transit agency and its ADA paratransit customers. She added that transit agencies that operate both ADA paratransit and microtransit should look into supporting both services with a single technology platform. What to Look for in a Technological Partner A problem that has plagued the paratransit industry since the advent of computerized scheduling has been, for many transit agencies, the lack of ongoing attention by software vendors. With the newer technologies potentially automating and changing traditional processes involved in para- transit scheduling and dispatching, it is imperative that the technology vendor be invested in the transit agency’s ability to successfully achieve its desired balance of service/cost efficiency and ser- vice quality. This type of relationship means that the vendor must be actively involved in the transi- tion from the old technology to the new technology and can help spot and solve issues with the data conversion or with the tablets or their mounts. The vendor must also be invested in the training of call and control center staff and drivers, that will patiently and systematically tune parameters until an initial optimal setting is reached, and that will be readily available—and has the bandwidth—to tweak the parameters further as the need arises throughout the licensing period. Transit agencies also need a vendor that can ensure that the data and reports that the transit agency needs are deliv- ered. In short, the technology partner must focus on the transit agency’s needs not only during the transition but also on an ongoing basis. In Sheboygan, WI, the Shoreline Metro director of transit and parking recommended that agen- cies make sure that the technology vendor provides great customer and technical support. Because any software or technology will stop working for any reason, the vendor should have contingency plans (e.g., multiple servers and web-based backup manifests). Shoreline Metro reported that its technology provider offers exceptional support and many layers of protection against inaccessibility to the system; the system has gone down just once in almost seven years. In Billings, MT, MET Transit’s transit supervisor advised looking for a technology partner with full CDO capabilities and a great track record of support, effective training, and responsiveness, especially in view of the learning curve for both dispatchers and operators. It is also important that transit agencies consider sustainability factors in looking for a tech- nology vendor. Does the technology vendor have sufficient bandwidth to support the transit

72 Continuous Dynamic Optimization: Impacts on ADA Paratransit Services agency? How have the vendor and its technology grown? For example, has the vendor proactively expanded its customer-support staff to adequately handle the needs of the transit agency, or has the vendor exhibited growth only after a contract has been signed, with newly hired staff still on a learning curve (when they are needed most)? What kind of ADA operational experience does the technology vendor have? And in terms of sustainability, is the vendor well financed for growth so as to provide the necessary future bandwidth with a well-stocked reserve for research and development? Is the vendor overly dependent on just one contract? Will the vendor even be there in five years? The answers to these questions are as important as the capabilities of the technology itself, and they contribute to the transit agency’s risk assessment in considering a new vendor/technology. Define Expectations and Goals, and Prepare to Adapt Shoreline Metro in Sheboygan, WI, advised transit agencies that are planning to implement or invoke CDO to “define (or redefine) your expectations for service. For example, be sure to train staff and drivers on how your software will work within the parameters you set up. Be sure to review current policies and match those with expectations and service standards.” While Shoreline Metro expected to become much more efficient and productive, the agency was not sure what the final service product would look like. It turned out that the agency was able to cut revenue vehicle hours by over 40 percent, which in turn drastically decreased the need for drivers and vehicles. Agencies that are interested in implementing CDO should be prepared to adapt by, for example, reducing resources, which might include reducing driver and staff revenue hours. Understand How the CDO Capability Works All 11 transit agencies—as well as the technology vendors interviewed—stressed how impor- tant it is for transit agency staff to understand the following: how the scheduling algorithms and parameters work in detail, how adjustments to a particular parameter or set of parameters affect the balance of service/cost efficiency and service quality in general, and how CDO can support an agency’s efforts to attain its goals. This is especially critical in the initial parameter tuning phase and on an ongoing basis. The transit agencies also agreed that while the iterative process of experimenting with parameter settings can be time-consuming, the end result in terms of finding that optimal setting is worth it. One of the technology vendors also warned against changing too many settings at once because a transit agency will not know which parameter change caused the result. In Summit County, UT, the High Valley Transit executive director also recommended having at least one staff person who understands the optimization algorithms and parameters in layman’s terms and is able to explain it to stakeholders (e.g., a board of trustees). The executive director said, “This is especially important if a contractor is employing CDO; the contractor should not be the only party that understands what is going on.” CDO May Result in a Changes in Staff Composition and Responsibilities Many of the transit agencies experienced changes in functions because CDO changes the way in which drivers and vehicles are controlled. Most of the agencies saw changes in the way that dispatchers do their job, with less focus on proactive dispatching because this process has been automated with CDO and more focus on addressing immediate issues that were not solved by the

Conclusions and Lessons Learned 73   CDO process. This shift often led to the repurposing of dispatch staff. Similarly, the use of CDO in some of the newer technologies to automate some of the more traditional scheduling func- tions has led to the repurposing of dispatch staff. In Sheboygan, WI, Shoreline Metro’s director of transit and parking also observed that “CDO is a whole different way of running operations and conducting business. It takes time to implement, and it will require patience while customers acclimate themselves to the service.” He further added that change is not necessarily bad but that staff can be stuck in their ways: “It should be remem- bered that not everyone in the organization will accept change.” In Wenatchee, WA, Link Transit’s general manager concurred, noting that his driver and dis- patchers were reluctant at first to embrace CDO. He added that the agency “had to limit drivers’ access to the next trip on their manifest, or drivers would not follow the CDO-generated manifest.” Eventually, the dispatchers’ perception they could do better subsided over time. Transit Agencies Still a Need a Dispatch Staff While CDO is able to solve problems and take advantage of opportunities to optimize the schedule, dispatchers are still needed for processing real-time problems such as no-show requests, incidents, accidents and breakdowns, requests for navigation assistance, and still- unscheduled trips. Regarding the last point, the CityLink general manager in Abilene, TX, observed that “CDO can still leave trips unscheduled if there is no room on the dedicated and overflow fleet. Consequently, dispatchers must constantly monitor this during peak times to avoid missed trips.” In Oxnard/Western Ventura County, CA, the Gold Coast Transit District manager of para- transit and special projects agreed, adding that its turnkey contractor’s proactive dispatching has been enhanced by its use of CDO. The technology “has not only had a positive impact on our OTP but also has freed the dispatcher to provide the human touch in customer service issues as they arise. In this way, CDO is directly and indirectly responsible for better supporting its most vulnerable population, who rely quite heavily on GO ACCESS for transportation.” CDO May Result in Changes to Operator Shifts and in Operator Turnover As a result of CDO, opportunities for changing run structures—to be more efficient—may arise. Agencies may wish to be cognizant, however, of other ramifications that might result from param- eter settings that favor service efficiency and productivity, depending on other circumstances. In Abilene, TX, for example, driver shortages at both CityLink and CityLink’s overflow provider meant that drivers were working 50- to 60-hour weeks. At the same time, CDO’s highly efficient runs were pushing drivers. CityLink management recognized that this combination, exacerbated by the extreme Texas temperatures in the summer, was causing driver fatigue because there was little recovery time. In response, CityLink shortened the work week by shortening the drivers’ straight shifts and converting some of the straight shifts to split shifts. In a related lesson learned in St. Petersburg, FL, the PSTA director of mobility services cautioned transit agencies to understand how the CDO technology handles operator lunch breaks. In Sheboygan, WI, the Shoreline Metro director of transit and parking further added, “Drivers will resign because they do not like the technology or because the technology is simply a new way of doing business. About 20 percent of drivers resigned or retired concurrent with CDO imple- mentation. However, with respect to the long term, new drivers were excited about the infused technology, and many agree that it makes their job easier.”

74 Continuous Dynamic Optimization: Impacts on ADA Paratransit Services Turn Off CDO if the Communication System or Tablets Are Malfunctioning If the drivers’ communication system is down or the tablets are malfunctioning, the senior director of demand response at CapMetro in Austin, TX, reminded transit agencies to turn off CDO immediately. Otherwise, a mismatch occurs because the changes that the CDO is making do not benefit from the actual vehicle location or the events as performed by the driver; nor are the CDO changes being communicated to the drivers’ tablets. Need for Additional Research Use of CDO Technology to Support Larger ADA Paratransit Service Models The authors of this synthesis found a dearth of cases in which CDO is being used in support of large ADA paratransit services. This is primarily because only a few technologies currently sup- port these very large ADA paratransit systems. Of these technologies, only one has demonstrated CDO capabilities, yet it is not being used because it requires a high degree of tailoring and experi- mentation before the transit agencies and their paratransit staff put their trust in the results. These large agencies appear to be holding back because of some combination of the following: they are content in trusting their dispatchers, they do not wish to invest in the tailoring/experimentation process (or they are contemplating a technology replacement), their technology vendor has been absent, or both. That said, the authors also see an increasing trend in the newer CDO technologies (that sprang from on-demand microtransit services) now being used to also support ADA paratransit services. Moreover, the size of the ADA paratransit services that these newer technologies are now sup- porting is increasing. For example, CapMetro in Austin, TX, which serves 2,000 ADA paratransit trips on an average weekday, is transitioning to one of these newer technologies to support its ADA paratransit services. As this trend continues, future research documenting the efficacy of these CDO technologies in supporting large ADA paratransit systems and more complex service models will be necessary. Use of One Technology Platform to Support ADA Paratransit, Microtransit, and Consolidated Services Many transit agencies are transitioning to the newer technologies to support their ADA para- transit services and their new or expanding microtransit services—that is, they are using one tech- nology platform for both. Some transit agencies are doing this because their older paratransit scheduling/dispatching technology is incapable of supporting the various forms of micro- transit. Thus, instead of keeping the current system and using a different technology to support microtransit services, the agencies’ preference is to have just one system capable of handling both services. In certain circumstances, such as in Green Bay, WI, the transit agency is now using one of the newer technologies to support both its ADA paratransit service and microtransit. While the two fleets operate separately, the drivers for both are ADA paratransit certified (employed by the operations/technology vendor) and can therefore be used to operate either service. In an emer- gency, a driver and vehicle from one fleet can be used for the other fleet. In Lubbock, TX, and Summit County, UT, both transit agencies have consolidated their ADA paratransit services and their microtransit services (Citibus in Lubbock has done so with its

Conclusions and Lessons Learned 75   human-agency-sponsored trips as well). Both agencies (Citibus and High Valley Transit) have been delighted with how the optimization technology supports ADA paratransit and microtransit, enabling them to combine the two services. As a result, ADA paratransit customers are using these microtransit services more and more. Just as TCRP Project B-48, “The Provision of Alternative Services by Transit Agencies: The Intersection of Regulation and Program Objectives,” is currently exploring the implications of ADA paratransit customers choosing to use a transit agency’s alternative services instead of the ADA paratransit service for certain trips, the same dynamic is worth exploring for these combined services. For example, the High Valley Transit executive director reported that the ADA paratransit portion of the total ridership in Summit County, UT, is actually very low (2 percent) in part because so many ADA paratransit customers are choosing on-demand trips via the microtransit component (while booking only advance trips and subscription trips with the ADA paratransit service.) In many service areas, ADA paratransit and microtransit are deployed to different locations because microtransit is commonly used in areas or at times when fixed-route service is not produc- tive or does not work for various reasons. But in some transit agencies, such as those in Lubbock and Summit counties, the two services overlap or are consolidated, and additional research should focus on these environments. Because these overlapping or consolidated on-demand services are now enabling ADA paratransit customers to get around on demand just as everybody else does, there is a need to research this dynamic as well as the extent to which CDO considers these different overlapping services in assigning trips.