Refuse trucks, commonly referred to as garbage trucks are a critical component of a municipality’s waste management industry. Their primary purpose is to collect, transport and deposit waste from households or businesses to designated transfer sites or dumps. Historically, refuse trucks have been powered by diesel fuel. The consumption of diesel fuel paired with the frequent accelerations or decelerations between each residential household along a route attribute to high amounts of tailpipe emissions and noise pollution within neighbourhoods. There is significant opportunity to explore avenues of powertrain electrification in refuse trucks to reduce their emissions and improve energy efficiency. To rapidly test promising powertrains, vehicle software models were developed. To accurately model the energy usage and power requirements of refuse trucks, environments for the models to operate were created. The environments were created using on-board diagnostic and positional data collected from refuse trucks in the City of Hamilton in Ontario, Canada. The data collection was done under a research collaboration between the City of Hamilton and the McMaster Automotive Resource Centre. The approaches used to develop the drive and duty cycles for the vehicle models offer some innovative approaches without the need for invasive devices to be installed. The powertrains that were modelled includes an all-electric, ranged extended electric and conventional refuse trucks. A comparative analysis of the pump-to-wheel powertrain efficiencies were completed looking at metrics such as fuel economy, payload capacity and fuel costs. Lastly, a look at truck emissions from a well-to-wheel perspective were completed to investigate the impact of each powertrain on greenhouse gasses and the effect on air quality of their immediate surroundings. / Thesis / Master of Applied Science (MASc)

This master‘s thesis pertains to the design of a garbage truck. The presented design offers a solution which respects the technical requirements of this vehicle, ergonomic needs of its crew and demands on the aesthetics of a modern vehicle for collecting and disposal of waste. The design cames from the knowledge gained in research part, historical, technical and design analysis. This work is generally set in to the present, so it respects today‘s technologies and manufacturing possibilities.

Ett konsortium bestående av Scania, JOAB, Powercell, KTH och Renova samarbetade för att designa och konstruera en bränslecellsdriven sopbil inom ett FFI-finansierade projekt. Sopbilen har sedan dess varit i drift i Göteborg från 2020 till 2023, med en vätgasinfrastruktur bestående av en tankstation vid tidpunkten för detta arbete. Under den tiden har bränslecellen genomgått kör- och stillastående tester. Verkliga kördata på sopbilens system och bränslecell registrerades. Databaserna var osynkroniserade i tid och därför krävdes datasynkronisering. Detta examensarbete inleddes med huvudsyftet att utveckla ett accelererat åldringstest för bränslecellen baserat på denna applikation. Ett ytterligare syfte var att utvärdera bränslecellens åldrande. På grund av de tillgängliga variablerna baserades bränslecellens åldrande på försämring av elektrisk effekt vid konstanta temperaturer och strömmar. En testcykel (eller effekt-cykel) baserad på testkörningen av sopbilen utvecklades istället. Genom att använda den etablerade metoden "k-means clustering" på bränslecellens effekt-cykler skapades en testcykel som var representativ för sopbils-körning från 2020 till 2023. Testcykeln validerades baserat på ett statistiskt kriterium, verifiering och ytterligare arbete krävs dock. Efter 141,80 timmars bränslecellsdrift kunde ingen åldring identifieras. Mer data från sopbilen behövs och faktumet att ytterligare en vätgastankstation kommer att installeras under 2023 i Göteborg innebär att sopbilens körmönster kan förändras. Resultaten från denna avhandling lägger dock grunden för framtida forskning och erbjuder ett tillvägagångssätt för att studera den bränslecellsdrivna sopbilen. / A consortium consisting of Scania, JOAB, Powercell Sweden AB, KTH, and Renova collaborated to design and engineer a fuel cell-powered refuse truck within a FFI-funded project. The refuse truck has been operational in Gothenburg since 2020, with a hydrogen gas infrastructure of one refuelling station at the time of this work. From 2020 to 2023, the fuel cell has gone through driving and standing still tests. Real drive data on the truck's system and fuel cell was recorded. The databases were unsynchronised in time, hence data synchronisation was required.  This thesis began with the main aim of developing an accelerated stress test for the fuel cell based on this application. Additionally, the aim was to evaluate the ageing of the fuel cell. Due to the available variables, fuel cell ageing was based on deterioration of fuel cell powers at constant temperatures and currents.  A test cycle (or power cycle) based on refuse truck test driving was developed instead. By utilising the established “k-means clustering” method on fuel cell power cycles, a test cycle representative of the truck operation from 2020 to 2023 was made. The test cycle was validated based on a statistical criterion, although verification and further work are required. After 141.80 hours of fuel cell power requests no ageing could be identified. More data from refuse truck operation is needed, also considering that an additional hydrogen refuelling station will be put in place in 2023 in Gothenburg, hence the drive pattern might vary. In this context, however, the results from this thesis lay the foundation for future research and offer an approach to study the fuel cell truck.

Indiana University-Purdue University Indianapolis (IUPUI) / This thesis proposes an artificial neural network (NN) model to predict fuel consumption in heavy vehicles. The model uses predictors derived from vehicle speed, mass, and road grade. These variables are readily available from telematics devices that are becoming an integral part of connected vehicles. The model predictors are aggregated over a fixed distance traveled (i.e., window) instead of fixed time interval. It was found that 1km windows is most appropriate for the vocations studied in this thesis. Two vocations were studied, refuse and delivery trucks. The proposed NN model was compared to two traditional models. The first is a parametric model similar to one found in the literature. The second is a linear regression model that uses the same features developed for the NN model. The confidence level of the models using these three methods were calculated in order to evaluate the models variances. It was found that the NN models produce lower point-wise error. However, the stability of the models are not as high as regression models. In order to improve the variance of the NN models, an ensemble based on the average of 5-fold models was created. Finally, the confidence level of each model is analyzed in order to understand how much error is expected from each model. The mean training error was used to correct the ensemble predictions for five K-Fold models. The ensemble K-fold model predictions are more reliable than the single NN and has lower confidence interval than both the parametric and regression models.

<div> This thesis proposes an artificial neural network (NN) model to predict fuel consumption in heavy vehicles. The model uses predictors derived from vehicle speed, mass, and road grade. These variables are readily available from telematics devices that are becoming an integral part of connected vehicles. The model predictors are aggregated over a fixed distance traveled (i.e., window) instead of fixed time interval. It was found that 1km windows is most appropriate for the vocations studied in this thesis. Two vocations were studied, refuse and delivery trucks.</div><div><br></div><div> The proposed NN model was compared to two traditional models. The first is a parametric model similar to one found in the literature. The second is a linear regression model that uses the same features developed for the NN model.</div><div><br></div><div> The confidence level of the models using these three methods were calculated in order to evaluate the models variances. It was found that the NN models produce lower point-wise error. However, the stability of the models are not as high as regression models. In order to improve the variance of the NN models, an ensemble based on the average of 5-fold models was created. </div><div><br></div><div> Finally, the confidence level of each model is analyzed in order to understand how much error is expected from each model. The mean training error was used to correct the ensemble predictions for five K-Fold models. The ensemble K-fold model predictions are more reliable than the single NN and has lower confidence interval than both the parametric and regression models.</div>