Hue City, the capital city of Thua Thien Hue Province located in central Vietnam, was selected as the study area (Figure 1). Hue City contains 27 wards with 354,124 people at a population density of 4779 persons per km² as of 2015 [21] . In 2014, the Vietnamese government approved “Decision No. 649/QD-TTg―Approval for Adjustments to General Planning for Hue City to 2030 and a Vision to 2050”, which called for Hue City to become environmentally sustainable [22] . Approximately 210 tons/day of waste is collected in Hue City and the general collection rates for the whole city and urban areas are approximately 89 and 90% - 95%, respectively [23] .

As a city with a UNESCO World Heritage Site, the number of tourists visiting Hue City rapidly increased at an annual growth rate of 10%, reaching 2.5 million in 2012. To fulfill the demands of the visitors, the number of accommodation

facilities also continually increased, reaching 402 facilities with 7762 rooms and 13,660 beds in 2012 [24] . Based on the guest house and hotel classification standards, the 402 facilities in Hue City are categorized into six classes: guest house (GH), 1-star hotel, 2-star hotel, 3-star hotel, 4-star hotel, and 5-star hotel [25] [26] .

To acquire representative samples of the hotel sector, the authors applied systematic sampling based on lists of hotels sorted by the number of beds. The lists were prepared separately for the abovementioned six classes, and the number of samples by hotel class is shown in Table 1.

The survey procedure followed the methodology presented by Matsui et al. (2015) [13] . The authors conducted three surveys for all target samples from June 2nd to 11th, 2012: actual waste measurement, waste composition and questionnaire surveys.

The authors requested the target facilities to keep their waste into the following three categories based on their typical separation manner:

− Separated recyclables (SRe): waste items separately kept for recycling, selling to the informal sector, or transferred to somewhere/someone else by the owners;

− Separated food residue (SFR): edible leftover food separately kept for feeding, collected by livestock breeders;

− General waste (GW): all remaining waste items collected daily by the formal waste collection sector, the Hue Urban Environment and Public Works State Company (HEPCO).

Source: General statistics office, 2015.

The authors assessed the waste separation rate at the target facilities based on their usual separation behavior, which was categorized into the following four patterns: 1) recyclables and food residue separation, 2) recyclables separation only, 3) food residue separation only, and 4) no separation.

Regarding the actual waste measurement survey, the surveyors daily visited all the target facilities and directly measured the amount of waste generated over 10 consecutive days. The first three days were spent preparing the surveyors and target facilities, and data from the following seven days were used for the analysis.

The waste composition survey was conducted to evaluate the recycling and composting potentials of GW. The authors selected 21 target facilities with recyclables and food residue separation. (The sample size by hotel class is shown in Table 6) GW was classified based on materials (plastic, paper, kitchen waste, rubber and leather, grass, textile, metal, glass, ceramic, and miscellaneous), types (container/packaging, product, and other), and their potentials for recycling and composting. The authors sorted GW into 10 physical categories and 77 detailed sub-categories as shown in Table 2, and regrouped GW into the following three sub-categories:

− Recycling potential (Re): the recyclable portion of the discharged GW, defined based on Hue City’s current informal sector trading market in 2012;

− Composting potential (Co): the compostable portion of the discharged GW, referred from the acceptable items established by Vietnamese composting facilities;

− Non-recyclable (NRe): The remaining portions of GW after the abovementioned recycling and composting potentials were considered.

The authors also conducted a questionnaire survey at the target facilities. The attributes and influencing factors on waste generation and recycling activities were collected and used for further analytical procedures.

The WGRs were calculated by dividing the daily amount of waste generation by four business scale indicators: number of rooms, number of beds, number of

*: Non-recyclable waste.

workers (including managers and serving workers), and number of guests as follows:

W G R i = D W A T N i (1)

where:

WGR_i: Waste generation rate of each target hotel by four indicators: waste generation amount per room, that per bed, that per worker, and that per guest;

DWA: Daily amount of waste generation of each target hotel;

TN_i: Total number of unit of each target hotel by four indicators: number of rooms, number of beds, number of workers, and number of guests.

To determine whether the WGRs followed a normal/Gaussian distribution, the authors applied a Shapiro-Wilk test [27] [28] and found that the WGR distributions were significantly different from a normal distribution (p < 0.001). Therefore, the authors employed non-parametric tests for further analysis.

In this study, the WGRs are reported as the mean with the 95% confidence interval (95% CI). Non-parametric bootstrap with replacement sampling method was applied to determine the 95% CI [12] . The authors assessed the difference in WGRs between hotel classes using a Kruskal-Wallis H test. And, a Mann-Whitney U test was used to identify significant differences between hotel classes by comparing all the pairs [28] .

The composition of GW was analyzed using the categories and sub-categories listed in Table 2, and the recycling and composting potentials were aggregated based on this table.

To establish the solid waste stream from the hotel sector in Hue City, the authors estimated the total amount of generated waste, as well as its components by hotel class based on the WGR, waste separation rate, and waste composition survey results collected in this study. The detailed procedure is introduced in Section 3 after these results are elaborated.

The waste separation rates of the target samples are summarized in Table 3 by the following 4 separation patterns: 1) recyclables and food residue separation, 2) recyclables separation only, 3) food residue separation only, and 4) no separation. The waste separation rates were calculated by dividing the number of targets applying specific waste separation activities by the total number of observed facilities by each hotel class.

W S R i j = N H S i j T N H j (2)

where:

WSR_ij: waste separation rate of separation pattern i of hotel class j;

NHS_ij: number of facilities with separation of separation pattern i of hotel class j;

TNH_j: total number of facilities of hotel class j.

The results show that most of the target hotels (43 of 45 facilities) separated their waste. All 3- to 5-star hotels separated recyclables and food residue, while the waste separation rates of GH and 1-star hotels were lower. This tendency seems to be consistent with that found in a study on the hotel sector in Cairo, Egypt, which reported that high-class hotels successfully sorted waste at its source [29] . A study on hotels in India by Malik, S.M. and Kumar (2012) also suggested that small hotels have not paid attention to solid waste management practices due to a lack of funds and knowledge, which may be a reason for the difference in the waste separation rate between hotel classes [30] .

WGRs are reported to differ between hotel classes, services, and regions [31] [32] [33] . A similar tendency was observed in a study on the hotel sector in Halong, Vietnam (Byer et al., 2006), which reported daily waste generation rates of 203.3, 114.3, and 89 kg/day at 4-, 3-, and 2-star hotels, respectively [9] .

Table 4 presents the total daily amounts of waste generated and WGRs (mean and 95% confidence interval) in three categories based on the hotel’s typical separation manner, calculated by the following four denominators: room, bed,

K-W test: H value by Kruskal-Wallis test by ranks; Spearman’s ρ: Rank correlation coefficient by Spearman method; *: p < 0.05; **: p < 0.01; ***: p < 0.001.

guest, and worker. Table 4 also shows the results of the Kruskal-Wallis H tests and Spearman’s rank correlation analyses.

Regarding the difference between hotel classes, the rank correlations between total daily amounts of waste generated and WGRs and room, bed, and guest were significantly positive (excluding the WGR of SRe per bed). The 5-star hotels exhibited the highest total daily waste generation (271.8 kg/day), followed by 4-star hotels (190.5 kg/day), 3-star hotels (68.8 kg/day), 2-star hotels (30.1 kg/day), 1-star hotels (7.8 kg/day), and GHs (5.3 kg/day). The daily waste generation and WGRs increased as the hotel increased. The results were consistent with those in past studies on the hotel sector [17] [31] [34] .

For the per-guest WGRs, 5-star hotels generated an average of 6.57 kg/guest/day, followed by 4-star hotels (2.32 kg/guest/day), 3-star hotels (1.71 kg/guest/day), 2-star hotels (0.48 kg/guest/day), 1-star hotels (0.60 kg/guest/day), and GHs (0.60 kg/guest/day).

According to some previous studies on the WGRs of the hotel sector in Asia [17] [35] [36] , the per-guest WGRs range from 0.8 to 3.33 kg/guest/day, which overlaps the range of 0.60 to 6.57 kg/guest/day presented in this study.

The per-worker WGRs of GW and SRe were not significantly rank correlated with hotel class, and those by GHs were highest. The number of workers employed by GH was generally much smaller than that employed by higher-class hotels, which could have caused the increased per-worker WGRs at GHs. The GW and SRe WGRs of 2- to 5-star hotels were similar, despite higher-class hotels employing greater numbers of workers. This could be attributed to the higher WGRs per room, bed, and guest exhibited by higher-class hotels.

The SFR WGRs was significantly rank correlated with hotel class. Most 3-5-star some 2-star, and a few 1-star hotels and GHs provide food and beverage services, such as restaurants and coffee shops/bars, which could be a possible reason for the higher SFR WGRs of higher-class hotels.

Based on the results of the Kruskal-Wallis H test, there were significant differences in the total per-room WGRs between hotel classes ( χ 2 ( 5 ) = 27.97 , p < 0.001 ) , and significant differences were also detected for other WGRs per bed, guest, and worker. To clarify the significant differences between hotel classes, a Mann-Whitney U test was conducted by comparing all WGR pairs. The results suggested that hotel classes could be re-grouped into: 1) Low (guest house and 1-star hotel); 2) Middle (2- and 3-star hotels); and 3) High (4- and 5-star hotels) classes. The WGRs were recalculated by these three hotel classes and presented in Table 5.

The authors also examined the influences of factors such as food service (breakfast, dining) and events (wedding party, conference) on WGRs. The Mann Whitney U test for each class indicated that there was no significant difference (p > 0.05) between hotels with and without food services/events.

Table 6 presents the physical composition (in percentage) of GW by hotel class. Food waste accounted for the largest proportion of GW for all hotel classes, and was highest for GHs (57.4%), followed by 5- (56.6%), 1- (51.0%), 3- (48.4%), 4- (44.1%), and 2-star hotels (40.9%). These results are similar to those from a study on ten hotels in Danang, Vietnam, by Otoma et al. (2013), who reported that food waste accounted for 65.5% of waste. Paper accounted for the second largest proportion, ranging from 10.1% (5-star hotel) to 23.9% (1-star hotel), followed by plastic, ranging from 10.7% (2-star hotel) to 15.5% (GH).

Table 6 also presents the waste composting and recycling potentials from GW. The composting potential was highest for most hotel classes, especially low-class hotels (53.2%). The recycling potential was highest for 2-star hotels, at 38.5%, followed by 4-star hotels (35.1%) and GHs (33.5%).

Plastic containers and packaging, paper containers and packaging, and paper products were the major three components that could be recycled, accounting for over half of the recycling potential for most hotel classes (excluding 2-star hotels). Plastic containers and packaging accounted for a major proportion of the recycling potential at low-class hotels (10.2%). In contrast, paper products

^[1]Low-class hotels: Guesthouse and 1 star hotels; ^[2]Middle-class hotels: 2 and 3 star hotels; ^[3]High-class hotels: 4 and 5 star hotels. K-W test: H value by Kruskal-Wallis test by ranks; Spearman’s ρ: Rank correlation coefficient by Spearman method; **: p < 0.01; ***: p < 0.001.

^aC & P: containers and packaging. ^[1]Low-class hotels: Guesthouse and 1 star hotels; ^[2]Middle-class hotels: 2 and 3 star hotels; ^[3]High-class hotels: 4 and 5 star hotels.

dominated the recycling potential at high-class hotels (11.5%). Although the hotel sector in Hue City conducts recycling using informal sectors such as junk-buyers and livestock breeders, substantial amounts of compostable and recyclable waste still remain in GW. Recycling activities tend to focus on materials that are easily separated and high value, such as beverage containers, while items that are not easily separated, such as a small amount of recyclables and smelly organic waste, are often put into GW without separation.

The estimation procedure followed the methodology reported by Matsui et al. (2018) [12] . To depict the solid waste stream from the hotel sector in Hue City, the authors estimated the total waste generation and its components by hotel class based on the per-room WGRs (Table 6), waste composition (Table 6), waste separation rates (Table 7), and total number of rooms (Table 7) by hotel class, as follows.

For the facilities separating both recyclables and food residue, the total waste generation could be divided into the following five components based on the waste’s composition and waste measurement survey data: recycling potential, composting potential, non-recoverable, separated recyclables, and separated food residue. For the facilities with other types of separation procedures, such as “recyclables only”, “food residue only”, and “no separation”, the authors estimated the amounts of waste generated for the five components using the waste composition data of facilities conducting “recyclables and food residue separation” according to assumed component allocations, as shown in Table 8. The authors assumed that the unseparated recyclables and food residue were discharged as a part of the recycling and composting potentials in GW at these facilities.

Based on the waste composition data of facilities separating “recyclables and food residue”, the total waste generation and its five components for each hotel class was calculated using the following equations:

^[a]Low-class hotel: Guesthouse and 1 star hotels; ^[b]Middle-class hotel: 2 and 3 star hotels; ^[c]High-class hotel: 4 and 5 star hotels. Source: General statistics office, 2015.

Recycling potential = n i ∗ [ ( a 1 + a 2 ) ∗ Re   + ( a 3 + a 4 ) ∗ ( Re   + SRe ) ]

Composting potential = n i ∗ [ ( a 1 + a 3 ) ∗ Co + ( a 2 + a 4 ) ∗ ( Co + SFR ) ]

Non-recyclable = n i ∗ ( a 1 + a 2 + a 3 + a 4 ) ∗ NRe

Separated recyclables = n i ∗ ( a 1 + a 2 ) ∗ SRe

Separated food residue = n i ∗ ( a 1 + a 3 ) ∗ SFR

Total waste generation I = Recycling potential + Composting potential + Non-recyclable + Separated recyclables + Separated food residue.

Where n_i is the total number of rooms in Hue City by each hotel class (Low-class, Middle-class, High-class).

To estimate the 95% CI of the total waste generation and its five components from the hotel sector in Hue City, the Monte Carlo simulation by non-parametric bootstrap method with return was applied [12] [37] .

The results of the total waste generation and its components by the Hue hotel sector are shown in Figure 2. The figure shows both the results of point estimation by the average value of each parameter and those of the 95% CI. From the point estimation results, the total waste generation of the hotel sector in Hue City was estimated to be 6.88 tons/day, 4.37 tons/day (64%) of which was GW, 2.13 tons/day (30%) was SFR, and 0.38 tons/day (6%) was SRe. The estimated composting and recycling potentials remaining in the GW were 2.57 (37%) and 0.87 tons/day (13%), respectively. The results show that the amount of waste treated at the landfill site could be reduced by 78%, from 4.37 to 0.94 tons/day, by improving waste separation at source.

High-class hotels contributed the highest amount of waste to the total amount generated by the hotel sector, at 3.24 tons/day (47%), followed by middle-class hotels at 1.98 tons/day (29%), and low-class hotels at 1.66 tons/day (24%). The recycling potential was highest for low-class hotels at 0.4 tons/day (6%), and the composting potential was highest for high-class hotels at 1.10 tons/day (16%).

The 95% CI of the total waste generation by the hotel sector was 6.28 - 7.62 tons/day, and the 95% CIs of the separated recyclable and separated food residue were 0.32 - 0.45 and 1.78 - 2.45 tons/day, respectively.

To estimate the impact of each parameter used for the confidence interval estimation of total waste generation, the authors also conducted sensitivity analysis by squaring the Spearman’s Rank Coefficients of each parameter, summing the results, and adjusting them to 100% [12] [38] [39] . The results presented in Figure 3 show that GW produced by high-class hotels contributed the most to sensitivity (38.2%), followed by GW produced by low-class hotels (22.3%), SFR produced by middle-class hotels (12.4%), and SFR produced by high-class hotels (11.7%). The GW produced by high-class hotels was the largest contributor, constituting 26% of the total waste generation. Therefore, it contributed the most to the sensitivity analysis result. To improve the reliability of the estimated total waste generation, the parameters with greater impacts on uncertainty require additional data collection and/or modeling by influence factors, which would reduce the overall uncertainty of the results.