The management strategy of pavement maintenance and repair problems for road networks has been the focus of prior studies over the past few years. Optimization approach has been a prominent method to solve such problems. The literature review is conducted in the following three topics: (1) the types of pavement distress; (2) RCPSP concept and applications; and (3) pavement maintenance and repair strategy.
2.1. Pavement Distress Types
The surface layer of flexible pavement construction uses asphalt mixtures as the major material. The next layer is the sub-base layer that particularly utilizes a hydraulically-bound mixture. The bottom layer, which is generally utilized unbound mixture material, is followed by the sub-grade and road-bed layer [
14]. The materials of the asphalt mixture include aggregates and bitumen. Aggregates provide a strong structure to sustain the vehicle loads and give skid resistance, wherein bitumen plays an important role to ensure the joint of the aggregates [
12]. Due to the exposure of the environment and traffic load, the pavement will endure distress and slowly lead to changes in the surface shape [
15].
Proper assessment on pavement conditions in road sections is essential in the pavement management strategy [
16,
17]. For the classification of defects found in flexible pavements, discrimination methods of defect types are usually utilized [
18]. Based on the distress identification manual from the Federal Highway Administration of the U.S. Department of Transportation, there are several types of asphalt concrete pavement distress, listed as follows [
19]:
Cracking, which is defined as fatigue damage caused by traffic volume exceeding the design standard or improper asphalt mix design and construction. A cracking shape is a line that has a gap, in which gap width is specified as less than 6 mm for low severity and more than 19 mm for high severity. There are six types of damage: fatigue cracking, block cracking, edge cracking, longitudinal cracking, reflection cracking at joints, and transverse cracking.
Patching and potholes, a portion of pavement surface. The dimensions of a damaged section are usually greater than 0.1 square-meter, and those damages are removed and replaced or additional material applied to the pavement after the original construction. For potholes, the pavement damage shape is bowl-shaped of various sizes in the pavement surface with a minimum dimension of 150 mm. The maximum pothole depth is below pavement surface, which includes high severity levels if the depth is greater than 50 mm.
Surface deformation is caused by poor material stability, changes in road base materials, or by braking and starting behaviors of vehicles, resulting in sagging or deformation in the horizontal and vertical directions at the place where the wheels are rolled over. This distress type creates a wavy surface with a width size similar to a vehicle wheel. There are two types of damage: rutting and shoving.
Surface defect is caused by the decomposition of granules, asphalt mud floating out of the surface, or the loss of bonding force between asphalt mud and granules. The physical feature of this distress is reflected in the reduction in pavement surface friction. This type of distress consists of three types: bleeding, polish aggregate, and raveling.
Miscellaneous distress is pavement damage that does not fall under the above three categories, mainly caused by segmental difference, which generates longitudinal slope irregularities at both ends of the structure. These distress types are included: lane-to-shoulder drop-off, water bleeding and pumping.
Potholes are the most general pavement distress, especially for asphalt concrete (AC) deterioration [
20,
21]. Before potholes appear, the distress commonly begins with pavement cracks. Pavement cracks occur because aging and environmental exposition chemically degrade the bitumen that binds pavement material [
22]. Through the cracks, water then penetrates the pavement and causes the loss of the adhesion between aggregate and binder. [
23]. The adhesion loss makes pavements further fragile. Thus, when the pavement experiences thermal exposure, fatigue, and vehicular burden, potholes then appear [
12,
15,
24]. If a small pothole is repaired as soon as possible, it will avoid further pavement deterioration and reduce repair cost. It can be proved that the repair delay after pothole appearance will give great chance for water penetrating the subgrade and cause further severe pavement structural failure [
9].
Based on distress classification types, a maintenance unit can perform a pavement distress survey to establish the composite condition index’s weight that reflects the pavement performance [
25]. Furthermore, according to the composite condition index’s value, a proper maintenance strategy can be established [
26]. The worst severity requires repair material and time to perform the repair activity. Thus, the severity level of pavement distress will influence the method and repair cost [
27].
2.2. Resource Constrained Project Scheduling Problem (RCPSP)
In the 21st century, many studies on resource-constrained project scheduling topics have generated numerous variants of optimization models with different considerations in resource constraints, activity assumption, and objective function. Resource project scheduling problem (RCPSP) has proven to be a reliable approach for solving project scheduling problems to minimize makespan with tight renewable resource constraints [
28]. The primary objective of RCPSP aims to minimize the total duration of the project schedule, which is composed of a set of activities connected by precedence constraints and restricted by multiple resource limitations [
29,
30,
31,
32].
Previous studies have proven that RCPSP can resolve complicated scheduling problems under the circumstances of complex resource allocation and assignment. A mixed-integer linear programming (MILP) optimization model was used in some previous studies to solve RCPSP, such as Pinto et al. [
33], who presented an extension of an RCPSP problem in a semiconductor industry multi-mode resource-collaboration and constrained scheduling problem (MRCCSP). Chakrabortty et al. [
34] presented a stochastic resource-constrained project scheduling problem (SRCPSP) model to minimize project makespan under stochastic activity durations. Hanzalek and Sucha [
35] considered a lacquer production scheduling problem into RCPSP with general temporal and resource constraints (i.e., positive and negative time-lags).
Other studies utilized a genetic algorithm (GA) technique to solve the RCPSP problem, such as Alcaraz and Maroto [
36], who presented a single-mode RCPSP environment model to achieve the minimum project duration. Rahman et al. [
37] considered three different practical environments—the information technology, the construction industry, and the medical service system—and attempted to examine the effectiveness of the proposed GA-based RCPSP model. Besides MILP and GA techniques, constraint programming (CP) techniques were also applied in previous studies. Wang et al. [
38] utilized CP techniques proposed a multimode resource-constrained project scheduling problem (MRCPSP) model combined with a work package-based information model in construction project scheduling. A CP solution model to solve resource-constrained multi-project scheduling problems that considered alternative activity chains and time flexibility (RCMPSP-ACTF) was conducted by Hauder et al. [
39].
2.3. Pavement Maintenance and Repair Strategy
Several previous studies attempted to solve the pavement maintenance problem through development of pavement maintenance and repair strategy. For example, a strategy for pavement resurfacing under steady-state conditions was proposed by Li and Madanat [
40] to optimize repair action’s frequency and intensity. Ouyang and Madanat [
41] introduced a mixed-integer nonlinear programming (MINLP) model to optimize highway pavement rehabilitation planning through minimizing the life-cycle cost in a finite horizon. Chootinan et al. [
42] proposed a methodology of multi-year pavement maintenance programming that explicitly accounts for unpredictability in pavement deterioration based on the genetic algorithm (GA) technique. Cohen and Madanat [
43] introduced an optimization model to minimize the total expected social cost of maintaining the facilities over a finite planning horizon under facility deterioration uncertainty.
Pamukovic et al. [
44] developed a decision support concept for maintaining damaged asphalt pavement by employing the multicriteria preference ranking organization method for enrichment of evaluation (PROMETHEE) method and the analytic hierarchy processing (AHP) method. Alcaraz et al. [
45] proposed a procedure to support collection, analysis, processing, and updating of pavement conditions to be implemented in pavement maintenance strategy based on a pavement management system. Mahanpoor et al. [
46] considered geometric road configuration and presented three particle swarm optimization (PSO) models to optimize the cost of pavement rehabilitation considering the existing pavement condition and project line elevation, travel time, as well as fuel consumption.
Biancardo et al. considered the infrastructure life cycle perspective and presented a decision support system (DSS) for maintenance operation management of runway pavement [
47] and roadway pavement [
48]. For the runway pavement, the DSS considered runway friction decay modeling based on aircraft traffic load. This study utilized classification and regression trees (CARTs), chi-squared automatic interaction detector (GCHAID),
k-means unsupervised learning algorithms, and Chiu’s subtractive clustering as data mining methods to process the aircraft traffic load data. The results of the data mining process were compared to the standard technical practice to evaluate the measurement value of runway friction decay. The DSS in roadway infrastructure maintenance operations management utilize a building information modeling (BIM) approach. The BIM model in this study used visual programming tools (VPL) that can inform the detailed technical information of pavement condition to the highway agency; thus, the highway agency can conduct proper maintenance actions.
Pavement repair involves machinery to produce, deliver, and spread asphalt concrete, activities which are associated with a large amount of resource consumption and greenhouse gas (GHG) emissions [
49]. Therefore, several studies attempted to develop a decision support system (DSS) for pavement maintenance strategy that considers environmental impact. Noland and Hanson [
50] examined life-cycle GHG emissions of asphalt concrete for pavements by adopting the green gas assessment spreadsheet for transportation capital project (GASCAP) method in a road maintenance project. A similar study conducted by Ma et al. [
51] was to investigate the GHG emissions of pavement maintenance during the life cycle period and compared sixteen different maintenance technologies of asphalt concrete. Yu et al. [
52] attempted to find optimal combinations of pavement performance, maintenance cost, and environmental impacts for developing asphalt pavement maintenance plans from the life cycle cost analysis (LCCA) perspective. Veropalumbo et al. [
53] compared mechanical and environmental performance of a base layer flexible pavement and considered the life cycle assessment (LCA) of bituminous mixture as an essential material of flexible pavement.
Further studies were focused on the LCA from a road network perspective with maintenance budget limitation. Chu and Huang [
54] introduced mixed-integer linear programming (MILP) models for network roads in three different pavement maintenance strategies: optimization-based, worst-first, best-first, and threshold-based from the life-cycle management perspective. Guo et al. [
55] proposed a probabilistic treatment path dependence (PTPD) model that considers pavement deterioration uncertainty to obtain the optimum maintenance budget. Mizutani et al. [
56] attempted to find an optimal global solution for the repair and work zone policy in the pavement management system at network level. Elhadidy et al. [
57] proposed a multi-objective pavement maintenance optimization model based on historical long-term pavement performance (LTTP) data to minimize the maintenance cost and maximize pavement condition.
Most of the above studies aimed to discuss pavement maintenance strategies from the deterioration model’s perspective, or predict the future pavement condition with life cycle analysis. Some of those studies considered establishing sustainable maintenance strategies in the long-term. However, only a few studies have attempted to develop pavement planned-repair strategies, such as Chen et al. [
58], who proposed an optimization model for a road network’s daily maintenance operations as a routing problem that considers service time’s uncertainty and operation cost. Lee [
59] introduced a scheduling model based on road users’ route-changing behavior to optimize contractor activities when repairing pavements. Huang and Lin [
10] proposed an arc routing problem approach to minimize the total cost of the construction machinery routing activity for road resurfacing projects on city road networks. Aarabi and Batta [
60] presented an optimization mode for repair worker routes on pothole repair problems in the road network. However, those pavement repair studies that aimed to obtain an optimum repair schedule utilizing the vehicle routing problem (VRP) approach for a problem-solving perspective.
The other approach to solve the pavement repair problem is from the perspective of the emergency repair concept. However, these types of studies have focused on network road repair after a natural hazard. Maya and Sorensen [
61] presented a road network repair strategy under scarce resources to maximize society accessibility in an affected disaster area. Based on the previous work, Maya et al. [
62] introduced the network repair crew scheduling and routing problem (NRCSRP) framework to minimize the traveling movement of a repair crew to recover an access road from the disaster point to the relief center. Moreover, Aksu and Ozdamar [
63] proposed a debris clearance scheduling model (DCSM), employing integer programming to recover blocked road network links in the first three days of emergency response under resource constraints. Inspired by their study, Kim et al. [
64] applied the ant colony system (ACS) algorithm as an optimization model for solving NRCSRP in the disaster response phase for short-term disasters; this optimization model was named as network repair crew scheduling for short-term disaster (NRCSSD). Moreover, most of those emergency repair studies also employed the VRP approach similarly to the pavement planned-repair studies above.
This study’s main contribution is to combine the pavement planned-repair strategy with the emergency repair concept to achieve a tangible schedule for immediate pavement pothole repair problems. Prior studies have mostly utilized the VRP approach to resolve the transportation problem and none of those studies have considered consumable resources in their models; this study strives to adopt the advantages of the RCPSP approach in handling consumable resource behavior on the proposed optimization model for pavement pothole repair problems.