Optimizing Forest Harvesting Efficiency: A Comparative Analysis of Small-Sized Logging Crews Using Cable-Grapple Skidders

Abstract

Examination of the technical and economic dimensions of skidding operations is imperative for sustainable forest management, offering invaluable insights crucial for the formulation of sustainable forestry strategies. In many countries, the shift from modified agricultural machinery to purpose-built forest machinery has become apparent in forest operations. However, this transition often accompanies a reduction in logging crew size, raising new questions about productivity, costs, and ergonomics of the introduced harvesting systems. This study investigates two skidding systems utilizing the cable-grapple skidder Welte 115/5L, differing in work team size: one with one skidder operator and two chainsaw operators (WT3) and the other with one skidder operator and one chainsaw operator (WT2). Conducted in natural European beech forests in southern Bulgaria, the research focused on the group shelterwood system within the Natura 2000 network. Both WT3 and WT2 exhibited net skidding productivity of 9.96 m³ PMH⁻¹ over a mean skidding distance of 300 m and a mean winching distance of 20 m, outperforming conventional systems in the area. Despite this, there were notable differences in gross skidding productivity (8.64 m³ SMH⁻¹ for WT3 vs. 7.30 m³ SMH⁻¹ for WT2), affecting skidding cost (EUR 5.41 m⁻³ for WT3 vs. EUR 6.62 m⁻³ for WT2) and unit production cost (EUR 9.33 m⁻³ vs. EUR 11.53 m⁻³). This study highlights that the cable-grapple skidder can be effectively employed by smaller teams, providing higher productivity, lower unit cost, and increased flexibility during piling, primarily due to the presence of the knuckle-boom loader. While WT2 experienced more delays and production pressure, the findings suggest that WT3 represents a balanced option for small logging crews, ensuring sustainable forest operations in the face of workforce challenges.

1. Introduction

In Bulgaria, forests account for about 4.5 million hectares, corresponding to 37% of the land area, and are mainly located in hill and mountain ranges [1]. Deciduous tree species predominate, among which European beech (Fagus sylvatica Linnaeus, 1753) and account for 70.5% of the total forest area and 55.5% of the total growing stock. During the last five years, the average annual timber production is in the range of 8.0–8.3 million m³ over bark (o.b.) which corresponds to 6.6–7.0 million m³ under bark (u.b.) [2]. As in many other European countries, the harvested timber volumes are significantly below the annual increment of the Bulgarian forests, and amount to 50–65% of the annual growth [3].

The conventional harvesting system in Bulgaria consists of motor-manual tree felling and processing, accompanied by an early stage of mechanized extraction, where skidders and yarders are used along with animals. The few operating harvesters are used in coniferous stands and their applicability in deciduous stands is very limited; in some cases, harvester operation is coupled with motor-manual processing, which cannot be considered a comprehensive technical-economical upgrade [4]. Timber extraction in the form of semi-suspended stems prevails due to the lack of forwarders. Almost 80% of the total harvested wood volume is extracted in Bulgaria by skidding, which is still the dominant extraction system in many countries (90% of the total harvest in Romania, 70% in Croatia and the Czech Republic) [4]. In this context, modified agricultural tractors represent the most widely used timber extraction equipment in Bulgaria, as well as in the Balkan countries, the Carpathians, and Italy [5,6,7,8,9]. In 2019, modified farm tractors represented 60% of the Bulgarian skidder fleet, and in northern Italy they still represent the largest majority thereof (>90%) [4,10]. However, modified agricultural tractors are inferior to purpose-built skidders in terms of mobility and productivity, and they often do not comply with the ergonomics and safety standards imposed on the logging industry [4]. Based on the steep terrain conditions, where most Bulgarian forests are located, it would be expected that cable skidders prevail, but this is not the case [10]. In fact, skidders may successfully replace modified farm tractors without requiring any substantial changes in the conventional harvesting methods [4], providing, at the same time, considerable upgrades in terms of productivity [11] and forest worker safety.

The total value of harvested wood products in Bulgaria increased by 270% between 2005 and 2017, from EUR 266 million to EUR 707 million. Nevertheless, in the same period, the forestry workforce was reduced by 9.8% from 12,000 to 10,700 [12], following a trend described for many European countries [13]. During recent years, the size of the logging teams has been reduced considerably from four to five members down to three or even two, including all forestry machinery and chainsaw operators. Given that this is a relatively new development, there is a knowledge gap in the productivity of such small-sized harvesting teams, especially in cases where the machine used is a cable-grapple skidder rather than a modified tractor.

The objective of this study was to examine the work cycle elements, productivity, and skidding costs of Welte 115/5L a cable skidder model equipped with a knuckle-boom loader, under two different harvesting team configurations. Team one consisted of one skidder operator and two chainsaw operators (WT3), while in team two there was one skidder operator working with one chainsaw operator (WT2). The interest in this research was if the smaller team size exerted a significant impact on productivity and costs in comparison with the larger team. This study was carried out during wood extraction in a European beech stand belonging to the Natura 2000 network that was managed according to the group shelterwood system. This paper aspires to shed some light on the productivity of small-sized harvesting teams equipped with a purpose-built skidder in times where the lack of forestry workers is becoming increasingly profound.

2. Materials and Methods

2.1. Study Site and Work Organization

This study was carried out in the Krumovgrad State Forest Range (

Table 1. Stand characteristics and skidding conditions in the study area.

Stand Characteristics	Value
Location	Gorni Yurutsi
Coordinates	N 41°19′08.79″; E 25°52′45.85″
Elevation	1080 m a.s.l.
Function	Natura 2000: BG 0001032, BG 0002019, habitat 9130
Species composition	European beech (Fagus sylvatica, L) 100%
Stand age	110 years
Stand type	High natural forest
Total harvested area	22.8 ha
Relative stocking	0.8
Logging operation	Group shelterwood cutting, thinning intensity 25%
Average tree height	22 m
Average tree DBH	36 cm
Average slope gradient	25° (47%)
Growing stock	361 m3 ha−1
Allowable cut	90 m3 ha−1
Main extraction direction	Downhill
Average slope gradient of skid road	9.7° (17.1%)

), which is located in the eastern Rhodope Mountains, near the Bulgarian–Greek border. It belongs to a Natura 2000 habitat type 9130 area, of which 22.8 ha were harvested. The average tree height was 22 m and the average tree diameter at breast height was 36 cm. The growing stock amounted to 361 m³ ha⁻¹ and the allowable cut to 90 m³ ha⁻¹.

The members of one harvesting crew, active in the study area, participated in the examined work team configurations: all three of them, one skidder operator and two chainsaw operators, had a minimum professional experience of at least 5 years and their ages ranged from 35–50 years. In both cases, the forest workers were not given any instructions to modify their working habits.

The marked trees were felled with chainsaws. All harvested wood was skidded in the form of semi-suspended stems. The skidding direction was downhill with the exception of a short uphill section. The average slope of the skid road was 9.7° (17.1%), which is typical for the area.

An articulated four-wheel-drive Welte 115/5L double-drum cable skidder (Welte Fahrzeugbau GmbH, Umkirch, Germany) equipped with a knuckle-boom loader was used (Figure 1,

Table 2. Technical specifications of the Welte 115/5L double-drum cable skidder.

Specification	Value
Engine power	84 kW at 2300 min−1
Length	6400 mm
Width	2500 mm
Height	3321 mm
Wheel base	3335 mm
Ground clearance	689 mm
Oscillation of front axle	±25°
Tires	4 × 23.1-26
Weight	7490 kg (unloaded)
Winch	Welte HZM-16, double drum, radio remote controlled
Cables length/diameter	90 m/13 mm
Tractive force	2 × 80 kN
Knuckle-boom	Welte SRK-6 with a round wood grapple for sorting and piling
Blade and shield	Front stacking blade; rear shield height-adjustable with roller support and clambunk

). The clambunk mounted on the skidder’s shield was not used during this study.

The WT3 logging crew consisted of three forest workers: the skidder operator, one chainsaw operator at the cutting area, responsible for tree felling and processing, and a second chainsaw operator, mostly located at the landing, unhooking and crosscutting the stems (Figure 2). At the beginning of the working day, the skidder operator also felled trees motor-manually for about 1–1.5 h before skidding. During this time, the skidder was not operating. Periodically, the chainsaw operator stationed at the landing would transition to the cutting area to carry out tree felling and processing tasks, as at the beginning of each shift. The skidder operator pulled out the main cable, hooked the logs to it, and controlled the winching process with the radio remote control of the winch.

The WT2 logging crew consisted of the skidder operator and one chainsaw operator. As in WT3, at the beginning of the working day, the skidder operator also felled trees motor-manually for about 3–4 h before skidding. That was necessary in order to facilitate a sufficient number of logs to be extracted later. During this time, the skidder was not operating. After three or four skidding cycles, the chainsaw operator moved to the landing and crosscut the skidded stems. The skidder operator did the same at the end of the shift.

2.2. Time Study

A time study was carried out to measure the duration of work elements and determine the productivity of the two examined harvesting systems, but its main focus was on analyzing the skidding operations. Tree felling and processing were regarded as a singular work element. This decision was taken so that erroneous results would be avoided, mainly due to the frequent moving of both work team members, especially of WT2, between the felling area and landing. However, time consumption per work team member for this unified work element was recorded at both sites.

A skidding work cycle was assumed to be composed of repetitive elements (Olsen et al., 1998 [14]). Each work cycle consisted of the following elements:

• travel empty (TE): traveling unloaded along the skidding trail;
• maneuvering (M): all necessary maneuvers for better machine positioning;
• hooking (H): time dedicated to pulling out the cable and hooking of the stems;
• winching (W): time consumed for pulling in the attached load to the skidder;
• travel loaded (TL): time consumed for the loaded travel of the skidder along the skid trail;
• unhooking logs (UH): time consumed for unhooking the logs;
• piling (P): time for picking up logs with the knuckle-boom loader and depositing them in large piles so that the logs were horizontal and parallel to each other, and their ends were lined up;
• delays (D): non-effective time due to organizational and technical delays.

During the study, information regarding the skidding distance, the winching distance, the average slope during each cycle, and the number and volume of the stems were also recorded. Each work cycle was measured by stopwatch and effective time was separated from delay time. Skidding distance and slope gradient of the skid roads were measured by a GPS. Winching distances and terrain slopes were measured with a professional laser rangefinder with an integrated clinometer. Individual log volume was determined by measuring the length and the mid-length diameter over bark, whereas the load volume consisted of the sum of all log volumes skidded together. Effective productivity values were calculated on net work times, whereas gross values included delays.

2.3. Costs

Costs were calculated using the COST model proposed by Ackerman et al. [15]. To determine the production and skidding cost for 1 m³ of timber, the cost analysis considered the following parameters: the number of operators, the hourly cost of an operator, the hourly cost of machines, the volume of the extracted timber, and the productive machine hours (excluding all delay times). The machine costs per hour were reported both as productive machine hours (PMH) excluding delays and scheduled machine hours (SMH). The purchase prices and operator wages required by the cost calculations were obtained from accounting records (Proto et al., 2016 [16]). Net labor remuneration was set to EUR 8.00 SMH⁻¹ for chainsaw operators and EUR 13.66 SMH⁻¹ for the skidder operator, exclusive of indirect salary costs (+52% in this case). Diesel fuel consumption was measured by refilling the chainsaw and skidder tanks and measuring the consumed fuel.

Cost calculations were based on the assumptions that forest companies in the region typically operate for an average of 150 working days per year and that a depreciation period of 10 years is reasonable. According to data collected from forest enterprises, wood extraction work is carried out for 130–150 working days per year, with a mean of six to seven effective workhours per day, assuming that one to two hours more are spent on justified delays daily (lunch times, resting pauses, machine maintenance) as well as unjustified delays. Thus, an annual total of 910–1050 SMHs with a utilization rate of 70% is to be expected [7,17,18].

2.4. Statistical Analysis

Data were analyzed with SPSS v.23 software. The General Linear Model (GLM) approach was applied to detect the effects of skidding cycles data on time consumption and productivity. The fixed factor examined was the work team size (WT) and continuous variables included the skidding distance (D_skid), the winching distance (D_winch), the number of the skidded stems (N_stems), the load volume (Volume), and the slope (Slope), which were treated as covariates. The general form of the GLM was (Equation (1)):

γ = μ + α_i(WT) + β₁ × D_skid + β₂ × D_winch + β₃ × N_stems + β₄ × Volume + β₅ × Slope + ε_ij

(1)

where μ = overall mean, α = fixed effect, β₁, β₂, β₃, β₄, and β₅ are constants, and ε = random error.

Initially, data were checked for outliers. Full factorial models were formed to examine possible interaction effects on the dependent variables. F-tests were conducted to examine the goodness-of-fit of the regression models and t-tests were used to test the significance of model coefficients. In a second step, insignificant factors were removed in order to create reduced models for predictive purposes of skidding time consumption, work cycle time consumption, and productivity. Validation of models’ normality and homoscedasticity was obtained graphically. The significance level for all tests was set at α = 0.05.

3. Results

A total of 76 work cycles, 38 for each work team, were carried out during which 331 stems with a total volume of 304.01 m³ were skidded and piled. Data collection lasted for 33.2 h, excluding all necessary preparation actions. No statistical differences were found between the two harvesting systems in terms of mean load volume (WT3: 4.02 m³, WT2: 3.98 m³) (t = −0.223, p = 0.814) and mean number of extracted stems per cycle (WT3: 4.34 stems, WT2: 4.37 stems) (t = 0.475, p = 0.636).

3.1. Productivity

In WT3, the skidder operator achieved a tree felling and processing productivity of 5.50 m³ PMH⁻¹, while the first and second chainsaw operators reached 5.39 m³ PMH⁻¹ and 5.25 m³ PMH⁻¹, respectively. It is important to note that these differences were not found to be statistically significant (F = 0.569, df = 2, p = 0.567). In WT2, the skidder operator achieved a tree felling and processing productivity of 5.26 m³ PMH⁻¹, while the chainsaw operator achieved 5.15 m³ PMH⁻¹. Similar to WT3, this difference was also not statistically significant (F = 0.477, df = 2, p = 0.491). However, it is worth highlighting that in WT3, only 18 trees (11% of the total tree count) were felled and processed by the skidder operator, compared to 60 trees (36%) in WT2. Also, delays accounted for 20% in WT3 compared to 34% in WT2. On average, in WT3, the tree felling and processing productivity per person was 5.35 m³ PMH⁻¹ or 4.28 m³ SMH⁻¹ compared to 5.20 m³ PMH⁻¹ or 3.43 m³ SMH⁻¹ in WT2.

The net productivity of both WT3 and WT2, obtained for a mean skidding distance of 300 m and a mean winching distance of 20 m was 9.96 m³ PMH⁻¹. However, gross productivity differed significantly between the two systems, reaching 8.64 m³ SMH⁻¹ in WT3 and 7.30 m³ SMH⁻¹ in WT2.

The mean values of most work cycle elements did not differ between the two work teams, except for the average work cycle delay time, which was 3.83 min in WT3 and 13.23 min in WT2. Consequently, the gross work cycle time for WT2 was higher than that of WT3 by 9.40 min on average (

Table 3. Descriptive statistics for the work cycle elements for the three-men work team (WT3) and the two-men work team (WT2).

Variable/Work Element (Unit)	WT3		WT2
	Mean Value ± SD *	Range	Mean Value ± SD *	Range
Travel empty (min)	4.68 a ± 3.94	1.65–12.75 12.75	5.74 a ± 3.28	2.42–12.75
Maneuvering (min)	0.42 a ± 0.12	0.17–0.75	0.43 a ± 0.12	0.17–0.75
Hooking (min)	2.04 a ± 0.68	0.9–3.22	2.16 a ± 0.65	1.07–3.68
Winching (min)	3.01 a ± 1.41	1.42–7.42	2.98 a ± 1.39	1.51–7.32
Travel loaded (min)	5.91 a ± 2.77	2.07–11.63	6.07 a ± 2.56	2.23–11.63
Unloading (min)	0.54 a ± 0.23	0.18–1.07	0.70 b ± 0.21	0.42–1.07
Piling (min)	8.06 a ± 2.30	0–11.63	7.40 a ±7.09	0–26.72
Delays (min)	3.83 a ± 2.91	1.83–14.73	13.23 b ± 22.13	1.58–78.92
Total cycle time (min)	28.48 a ± 7.19	11.88–42.07	38.74 b ± 20.95	15.4–94.63
Delay-free cycle time (min)	24.65 a ± 7.35	9.58–40.07	25.51 a ± 8.91	10.73–49.75
Number of stems (n)	4.34 a ± 0.48	4–5	4.37 a ± 0.49	4–5
Load volume (m3)	4.02 a ± 0.52	2.6–4.85	3.98 a ± 0.36	3.1–4.75
Net productivity (m3 PMH−1) *	10.70 a ± 3.07	6.14–25.67	10.43 a ± 3.71	4.69–23.76
Gross productivity (m3 SMH−1) *	9.11 a ± 4.52	5.39–21.71	7.50 b ± 2.92	2.38–14.81
Empty turn (m)	207.68 a ± 164.82	75–510	288.89 b ± 132.19	120–530
Skidding distance (m)	269.29 a ± 142.57	75–510	289.03 a ± 132.07	120–530
Winching distance (m)	17.61 a ± 5.65	8–30	16.97 a ± 4.69	8–25
Number of cycles per SMH *	2.28 a ± 0.77	1.43–5.05	1.89 b ± 0.73	0.63–3.9
Slope gradient of skid road (°)	9.68 a ± 1.53	7–11	10.16 a ± 1.41	8–13
Mean cycle speed (km h−1)	2.67 a ± 0.22	2.15–3.04	2.98 a ± 0.43	2.43–5.03
Mean speed loaded (km h−1)	2.65 a ± 0.36	2.2–3.23	2.89 a ± 0.74	2.11–6.4
Mean speed unloaded (km h−1)	2.74 a ± 0.26	1.95–5.33	3.18 b ± 0.50	2.19–4.32

Note: Different letters denote statistically significant differences among the means at the α = 0.05 level; * SD: standard deviation, PMH: productive machine hour, SMH: scheduled machine hour.

).

The breakdown by main groups of work cycle elements in delay-free cycle time (

Table 4. Distribution of the work element times during the skidding operations as a percentage of the net time (^a) and the gross time (^b) in the examined harvesting systems.

Work Element	Work Team
	WT3 a (%)	WT2 a (%)	WT3 b (%)	WT2 b (%)
Travel empty	29.3	22.5	27.5	14.8
Maneuvering	1.1	1.7	1	1.1
Hooking	6	8.5	5.6	5.6
Inhaul	7.3	11.8	6.9	7.8
Travel loaded	32.6	23.8	30.7	15.7
Unload	1.5	2.8	1.5	1.8
Piling	22.2	29	20.9	19.1
Organizational delays	-	-	2	30.1
Technical delays	-	-	3.9	4

) shows the predominance of the travel operations of the skidder (TE and TL) with the largest share of 61.9% in WT3 and 46.3% in WT2, followed by piling (22.2% in WT3 vs. 29% in WT2). Load attachment (H and W) had a share of 14.4% in WT3 and 22% in WT2, whereas unloading times were the lowest, 1.5% in WT3 and 2.8% in WT2.

One considerable difference between the two harvesting systems during the skidding operations is the delay time percentage, which amounted for only 6% of the total work time in WT3 compared to 34.1% in WT2 (Table 4). In the first case, the majority of delays (3.9%) were due to mechanical delays, whereas 30.1% of them in WT2 were due to organizational reasons, most notably waiting for the felling of trees in the cutting area.

The mean speed of the studied skidder was 2.67 km h⁻¹ in WT3 and 2.98 km h⁻¹ in WT2, and 2.65 km h⁻¹ and 2.89 km h⁻¹ for loaded vehicles, respectively. Significant differences were found only in the mean speed of the unloaded skidder amounting to 2.74 km h⁻¹ in WT3 and 3.18 km h⁻¹ in WT2, which could be attributed to differences in terms of terrain and skid road conditions.

3.2. Costs

The net costs for downhill skidding with the Welte 115/5L cable skidder were calculated at EUR 57.73 PMH⁻¹ and EUR 68.83 PMH⁻¹ for WT3 and WT2, respectively (

Table 5. Skidding cost analysis for the examined systems.

Cost Item	WT3		WT2
	Chainsaw	Skidder	Chainsaw	Skidder
Investment (EUR)	1250	60,000	1250	60,000
Interest rate (%)	7	7	7	7
Expected economic life (years)	5	10	5	10
Utilization (PMH year−1)	540	1152	540	1152
Depreciation (EUR year−1)	243	5184	243	3888
Interest (EUR year−1)	56.63	2491.44	56.63	2446.08
Fuel cost (EUR PMH−1)	2.98	12.2	2.98	12.2
Lubricants (EUR PMH−1)	0.45	1.83	0.45	1.83
Maintenance and repair (EUR MH−1)	0.35	4.2	0.35	4.2
Net cost incl. worker remuneration (EUR PMH−1)	20.96	57.73	26.26	68.83
Net skidding cost (EUR m−3)	3.92	5.41	4.91	6.62

). Nevertheless, due to the different productivity rates, the net skidding cost per production unit was EUR 5.41 m⁻³ in WT3, and 22% higher or EUR 6.62 m⁻³ in WT2 (Figure 3).

Additionally, the tree felling and processing cost has been calculated at EUR 3.92 m⁻³ and EUR 4.91 m⁻³ for the WT3 and WT2, respectively. Therefore, the production cost totaled EUR 9.33 m⁻³ for WT3 and EUR 11.53 m⁻³ for WT2.

3.3. GLM Analysis of Variance and Prediction Models

The main factors affecting the effective skidding cycle time (Model 1) were the skidding distance, the winching distance, the load volume, and the work team size (

Table 6. Analysis of the variance table for the GLM models developed for the skidding operations.

Model	Dependent Variable:	Source	SS	df	F	p-Value	Partial η2
1	Net skidding cycle time	Corrected Model	2450.445	6	86.572	0.000	0.883
		Intercept	6.288	1	1.333	0.252	0.019
		Skidding distance	1251.138	1	265.211	0.000	0.794
		Winching distance	91.097	1	19.310	0.000	0.219
		Load volume	31.440	1	6.664	0.012	0.088
		Work team size	19.867	1	4.211	0.044	0.058
		Error	325.509	69
2	Net work cycle time	Corrected Model	2719.851	6	13.994	0.000	0.549
		Intercept	177.066	1	5.466	0.022	0.073
		Skidding distance	1642.075	1	50.690	0.000	0.424
		Error	2235.199	69
3	Gross work cycle time	Corrected Model	5499.118	6	4.316	0.001	0.273
		Intercept	0.463	1	0.002	0.963	0.000
		Skidding distance	1282.155	1	6.038	0.017	0.080
		Number of stems	1305.366	1	6.147	0.016	0.082
		Work team size	1399.380	1	6.590	0.012	0.087
		Error	14,652.337	69
4	Net productivity	Corrected Model	517.552	6	11.733	0.000	0.505
		Intercept	0.006	1	0.001	0.977	0.000
		Skidding distance	248.422	1	33.791	0.000	0.329
		Load volume	139.591	1	18.987	0.000	0.216
		Slope	40.396	1	5.495	0.022	0.074
		Error	507.272	69
5	Gross productivity	Corrected Model	308.081	6	8.752	0.000	0.432
		Intercept	14.449	1	2.463	0.121	0.034
		Skidding distance	90.146	1	15.366	0.000	0.182
		Load volume	160.816	1	27.412	0.000	0.284
		Number of stems	70.546	1	12.025	0.001	0.148
		Work team size	30.199	1	5.148	0.026	0.069
		Error	404.793	69

and

Table 7. Goodness of fit and parameter estimates for the general linear models of the study.

Model	Adjusted R2	Parameter	B	s.e.	t-Value	p-Value	95% C.I.		Partial η2
							Lower Bound	Upper Bound
1	0.873	Intercept	4.586	3.544	1.294	0.200	−2.484	11.656	0.024
		Skidding distance	0.036	0.002	16.285	0.000	0.032	0.041	0.794
		Winching distance	0.275	0.063	4.394	0.000	0.150	0.400	0.219
		Load volume	−1.742	0.675	−2.582	0.012	−3.089	−0.396	0.088
		(Work team size = 3)	−1.040	0.507	−2.052	0.044	−2.052	−0.029	0.058
		(Work team size = 2)	0
2	0.510	Intercept	21.852	9.287	2.353	0.021	3.325	40.378	0.074
		Skidding distance	0.041	0.006	7.120	0.000	0.030	0.053	0.424
3	0.210	Intercept	3.263	23.777	0.137	0.891	−44.171	50.697	0.000
		Skidding distance	0.037	0.015	2.457	0.017	0.007	0.066	0.080
		Number of stems	10.056	4.056	2.479	0.016	1.965	18.146	0.082
		(Work team size = 3)	−8.733	3.402	−2.567	0.012	−15.519	−1.946	0.087
		(Work team size = 2)	0
4	0.462	Intercept	−0.209	4.424	−0.047	0.962	−9.035	8.617	0.000
		Skidding distance	−0.016	0.003	−5.813	0.000	−0.022	−0.011	0.329
		Load volume	3.671	0.843	4.357	0.000	1.990	5.352	0.216
		Slope	0.556	0.237	2.344	0.022	0.083	1.029	0.074
5	0.383	Intercept	5.522	3.952	1.397	0.167	−2.362	13.406	0.028
		Skidding distance	−0.010	0.002	−3.920	0.000	−0.015	−0.005	0.182
		Load volume	3.940	0.753	5.236	0.000	2.439	5.442	0.284
		Number of stems	−2.338	0.674	−3.468	0.001	−3.682	−0.993	0.148
		(Work team size = 3)	1.283	0.565	2.269	0.026	0.155	2.411	0.069
		(Work team size = 2)	0

). On the contrary, the effective work cycle time (Model 2), which also included the work element of piling, was affected only by the skidding distance. Skidding distance, the number of stems per load, and the work team size exerted a significant effect on the gross work cycle time (Model 3). It should be noted that a gross work cycle lasted, on average, 8.7 min more in WT2 than it did in WT3. In the case of net productivity, its magnitude was largely determined by the skidding distance, load volume, and slope (Model 4). Finally, gross productivity was affected by skidding distance, load volume, number of stems, and the work team size (Model 5).

4. Discussion

4.1. Tree Felling and Processing Productivity

The tree felling and processing productivity of both systems per working person was in the range of 5.20–5.35 m³ PMH⁻¹. When delays were included, the gross productivity rates decreased to 4.28 m³ SMH⁻¹ and 3.43 m³ SMH⁻¹, respectively. Latterini et al. [19] have reported a wide range of motor-manual productivity rates in beech stands, starting from as low as 0.40 m³ SMH⁻¹ to as high as 20.6 m³ SMH⁻¹. This wide variability may be attributed to many factors, including the harvesting system applied [20], but, most importantly, to the average DBH of the felled trees [21,22]. Additionally, it is important to highlight that WT2 was anticipated to operate at a notably slower pace compared to WT3 when it comes to processing and delivering the same quantity of felled and processed wood products. Consequently, in scenarios such as salvage logging, logging crews as small as that of WT2 may struggle to efficiently and swiftly process the required volume of wood products. This inefficiency is primarily a result of WT2′s smaller size, which can hinder its ability to match the productivity and output of larger logging crews like WT3.

4.2. Skidding Productivity

The net skidding productivity of both WT3 and WT2, obtained for a mean skidding distance of 300 m and a mean winching distance of 20 m, was 9.96 m³ PMH⁻¹. Nevertheless, there was a notable disparity in gross productivity between the two systems, with WT3 achieving 8.64 m³ SMH⁻¹ and WT2 achieving 7.30 m³ SMH⁻¹. This result may be attributed to the considerable amount of delays recorded in WT2, accounting for 34.1% of the total work time, compared to only 4.1% reported for WT3. Organizational delays including, in most cases, the frequent relocation of the forest workers and switching between tasks were required by WT2 as the volume of harvested and processed wood was not enough to keep the skidding undisturbed. Behjou et al. [23] reached the same conclusion after examining skidding productivity in the Caspian forests of Iran, and suggested the careful planning and execution of the forest operations as a solution, especially in sensitive sites.

A second and very promising finding refers to the obtained skidding productivity rates for both WT3 and WT2. It should be noted that skidding productivity rates of this magnitude in the specific region are twice as high as those achieved with modified agricultural tractors with larger logging teams, usually consisting of four or five workers. This is a first indication that the combination of logging crews of smaller size than conventional ones, when combined with a purpose-built machine such as the examined grapple skidder, can lead to considerable productivity increases over the widely used modified agricultural tractors. Specialized machinery has distinct advantages; however, the lack of capital is crucial for forest entrepreneurs who wish to minimize the associated financial risk [24]. In this context, alternative skidding equipment [4] could be considered and tested under the Bulgarian conditions in the future.

Skidding distance, load volume, and number of stems, often combined, have been frequently identified as factors exerting a significant impact on time consumption during skidding operations. Skidding distance has been a consistent component of all productivity- and cycle time-related models in the present study, as well as in previous research (e.g., [7,25,26,27]). Load volume was found to impact effective skidding cycle time (Model 1), net productivity (Model 4), and gross productivity (Model 5), which is a finding also consistent with the findings of Orlovský et al. [25]. Finally, the number of stems affected the gross cycle time (Model 3) and gross productivity (Model 5), but did not influence effective skidding cycle time or net productivity. This finding differs from previous research on skidding productivity, such as the studies by Mousavi et al. [28], who exhibited the dependence of the overall time consumption of the HSM-904 skidder work cycle on the number of logs and the load volume, and by Najafi et al. [29], who showed that the skidding time consumed during the operation of the HSM-904 skidder depends on the skidding distance and the number of logs in the load. Our finding may be attributed to two factors: (i) The almost identical number of stems per skidding cycle in the two harvesting systems (4.34 stems in WT3 vs. 4.32 stems in WT2) and (ii) the extensive experience of the skidder operator. Consequently, when the skidder operator worked without interruptions, as in WT3, the number of stems had a limited impact on the effective cycle time or net productivity, unlike in WT2 where delays increased.

The productivity rates in the present study are higher than those reported by Orlovský et al. [25] for a LKT 81ILT skidder equipped with a knuckle-boom loader. In that study, the mean skidding distance was 316 m and the mean load volume per cycle was 8.01 m³, resulting in 6.31 m³ PMH⁻¹ and 4.21 m³ SMH⁻¹, respectively. A variant of the same machine (LKT 81T) which was not equipped with a knuckle-boom loader was examined by Stoilov and Krumov [30], who obtained a productivity of 6.27 m³ PMH⁻¹ during the extraction of beech stem sections at an average skidding distance of 1290 m and a load volume of 5.65 m³. In that study, the skidder did not pile the logs and the harvesting team consisted of five forest workers. Borz et al. [6] reported for the TAF 690 OP skidder net and gross productivity rates of 4.41 m³ PMH⁻¹ and 3.12 m³ SMH⁻¹, respectively. These rates were obtained for a shorter average winching distance of 8.7 m, a four-times longer average skidding distance of 1706.3 m, a load volume of 4.89 m³, and 6.48 stems per turn. For the more powerful and heavier John Deere 548H skidder than the one used in our study, Proto et al. [7] reported average values of seven stems extracted per cycle, 276 m of skidding distance (range 34–65 m), and volume per turn of 3.88 m³, which resulted in skidding productivities of 30.4 m³ PMH⁻¹ and 24.8 m³ SMH⁻¹. These values are considerably higher than those in our case and may be attributed to the much shorter skidding distances. Behjou [27], after examining beech wood extraction in Iran, reached the same conclusion and reported a productivity of 34.97 m³ PMH⁻¹ declining to 11.22 m³ PMH⁻¹ for the skidding distances of 50 m and 425 m, respectively, with the rate of productivity decline being higher for lower skidding distances.

In similar studies, slope has been identified as a factor affecting the productivity of skidding operations [23,31]. We found a relationship between slope and net productivity (Model 4) but the impact was rather limited. This result, however, may be due to the experiment setup, and our focus on the impact of work team size. Generally, when working on steep terrain and sensitive sites, such as the study area, harvesting and skidding operations should be carefully planned and executed [23].

4.3. Cost

The reported skidding cost of EUR 5.41 m⁻³ for WT3 is comparable to that of similar studies, whereas the cost of EUR 6.62 m⁻³ for WT2 is among the highest. However, these results are not comparable as piling was also included in the present study. Horvat et al. (2007) reported costs for the Ecotrac 120 V cable skidder of EUR 4.88 m⁻³ for a hilly working site and EUR 5.25 m⁻³ for a mountain working site with a skidding distance of 300 m. Proto et al. [7] reported extraction costs of EUR 4.50 m⁻³ and EUR 3.90 m⁻³ while using the winch and the grapple of John Deere 548H (Deere & Company, Moline, IL, USA), respectively. According to Lotfalian et al. [31], the unit cost for Timberjack 450C in the Caspian region, considering the gross and net production rate, was EUR 4.7 m⁻³ and EUR 5.7 m⁻³, respectively.

Due to the decreased work team size, the total cost per PMH for WT3 was higher than that of WT2 (EUR 99.65 vs. EUR 95.09 PMH⁻¹). Nevertheless, this economic advantage was not the case for the cost per production unit; due to the large amount of delays, WT2 had a 23.6% higher production cost (EUR 11.53 m⁻³ in WT2 compared to EUR 9.33 m⁻³ in WT3). More specifically, this is partially due to the better-paid skidder operators (EUR 13.66 SMH⁻¹) that are requested to do the less-paid tasks of chainsaw operators (EUR 8.00 SMH⁻¹) in WT2. This might prove a friction point in work team dynamics in the long run. Furthermore, the reduction in the work team size has been found to affect the quality of skidding operations. Working under increasing pressure due to large delays resembles the pressure to reach production standards [32,33]. Psychological pressure can affect cognitive abilities (i.e., perception and problem solving) through affecting the workers’ mental reserves [34,35] and result in lesser quality of forest operations, both in terms of the produced timber assortments and the environmental impacts of forest operations [36]. Similarly, stressful situations may create adverse working conditions, leading to increased frequencies of accidents and musculoskeletal disorders [37,38].

4.4. Travel Speed

Our findings indicate empty and loaded travel speeds of approximately 3.00 km h⁻¹ (speed range 2.65–2.71 km h⁻¹ in WT3 and 2.89–3.18 km h⁻¹ in WT2). The slightly higher speed in WT2 may be attributed to the pressure for more productivity exerted on the skidder operator. Orlovský et al. [25] reported that for the LKT 81 ILT cable skidder, also fitted with a knuckle-boom, mean speeds with and without load of 3.75 km h⁻¹ and 4.45 km h⁻¹ were respectively, relatively close to those we found. However, Spinelli and Magagnotti [4] reported empty and loaded travel speeds for a 96 kW agricultural tractor of 8.1 and 7.3 km h⁻¹, respectively, which are considerably higher. Proto et al. [7] also reported higher average speeds for a John Deere skidder 548H, 8.58 km h⁻¹ for unloaded travel and 6.02 km h⁻¹ for loaded travel. In theory, it is possible to reduce the time required for wood extraction by increasing the travel speed during both loaded and unloaded turns. However, it is important to note that the terrain conditions in our study do not allow for a substantial increase in travel speed.

4.5. Optimizing Logging: Mechanization and Workforce Training

With the declining number of employees in the logging industry not only in Bulgaria but also in many other countries, due to the unfavorable demographic changes, the issues of increasing the share of mechanized operations and introducing higher-performance logging equipment arise. Mechanization allows for higher operator productivity, and for this reason achieves an advantage to traditional technology, which makes it a more attractive choice even when utilization rates and labor cost are comparatively low [18]. However, mechanization efforts should not be limited to skidding, but rather include all aspects of forest operations. Slugeň et al. [39] determined the productivity of a John Deere 1070D harvester in thinning operations of Sessile oak (Quercus petraea Liebl.) and European beech (Fagus sylvatica L.) coppice stands of over 50 years of age to be 6.36 m³ PMH⁻¹ (5.35 m³ SMH⁻¹). Lower productivity was attributed to the numerous and thick branches, combined with the frequent presence of trees with multiple treetops. Future technological developments may increase the efficiency of mechanized harvesting of hardwood species and provide a more feasible alternative to the conventional motor-manual wood harvesting [40].

Special attention should be given to the forest workforce. If the trend of smaller-sized work teams continues, the need for specialized training courses adjusted to the characteristics and needs of individual groups of workers (Poje et al., 2016) is expected. According to our findings, the combination of a dedicated skidder and two chainsaw operators has been found to be more efficient than the conventional larger harvesting and skidding teams combined with modified general-purpose machinery, mostly tractors. This is important and should be communicated to local forest entrepreneurs since even smaller logging teams, as in the case of WT2, should be avoided. Training courses focused on communication and teamwork seem to be of special relevance in the study area, given the move from larger to smaller logging teams. Additionally, training courses can help workers develop entrepreneurial skills, increasing their productivity by introducing more efficient working methods and equipment [41,42,43], and additionally reducing the environmental impacts of forest operations [36]. The investigated cable skidder, thanks to the knuckle-boom loader, can also be used to pick up logs and bolts and deposit them in large piles, a requirement for further transport. Such possibilities increase the versatility of forest entrepreneurs and could result in higher adoption of similar purpose-built equipment in the coming years. The examined cable-grapple skidder may provide a better alternative, in terms of economic and technical efficiency, to the conventional logging teams in southern Bulgaria, where modified agricultural tractors prevail. More future research is needed to elaborate this point, as the forest workforce is declining countrywide and small-sized logging crews are more frequent than ever. The promotion of forest mechanization in the area can be facilitated by the acquisition of second-hand equipment from abroad; however, it will be suboptimal unless the local workforce is trained in modern work methods and equipment.

4.6. Limitations and Strengths of this Study

A limitation of the present study pertains to the absence of a detailed analysis concerning tree felling and processing operations. A more in-depth examination of these work elements could have yielded additional data and enhanced insights into work team dynamics, thereby providing a more comprehensive understanding of both harvesting systems. Unfortunately, this detailed analysis was unfeasible due to budgetary constraints. Consequently, the focus of the study shifted to the skidding operations. Despite this limitation, this study offers valuable productivity and cost data for all operations, thereby contributing to a comprehensive overview of the situation. Additionally, this study addresses the impact of small logging crew size on productivity and cost, a topic that has not been extensively explored and is anticipated to gain significance in the future as a potential threat to the sustainability of forest operations.

5. Conclusions

Both studied harvesting systems WT3 and WT2 outperformed the existing ones, achieving the much higher productivity rates of 9.96 m³ PMH⁻¹, twice higher than those of the commonly implemented harvesting systems in this region that use modified agricultural tractors for skidding with work teams that consist of more chainsaw operators.

Despite the productivity gains, the transition from WT3 to WT2 resulted in large organizational delays that reduced the gross productivity in WT2 to 7.5 m³ SMH⁻¹ compared to 9.11 m³ SMH⁻¹ in WT3. Furthermore, WT2 was more stressful for the work team members, who had to move more frequently between the stand and the landing to deal with irregular flows of stems to be skidded. Thus, WT3, comprising one dedicated machine and two chainsaw operators, seems to be a considerably more balanced and efficient option for the study area compared to WT2, with delays accounting for only 4% of the total work time. To make optimal use of such smaller-sized work teams, special attention should be given to the training of the existing work force on modern work methods and equipment, especially in times of large cutbacks in the forestry workforce.