Review of Root Intrusions by Street Trees and Utilising Predictive Analytics to Improve Water Utility Maintenance Strategies

Abstract

Tree root intrusion can cause failures of underground sewer pipes and thus represent a major water asset management issue. If tree root intrusion is not detected early, this may lead to the interruption of wastewater services and high costs of repair to the pipeline. The objectives of this review are to assess the existing maintenance strategies, explore suitable strategies for Australia and similar settings around the world, and identify possible factors and predictive tools. Maintenance strategies can be divided into two categories: reactive and proactive approaches. The current reactive approaches are (1) mechanical techniques to clean the root mass in pipe networks and (2) chemical techniques to remove the root mass and control future growth. The literature suggests that the reactive approaches often provide only partial solutions. The proactive approaches, guided by a predictive model of tree root intrusion and its related factors, showed the potential to improve maintenance and limit the risk of the damage from re-occurring. Predictive models could help to evaluate the risk of planting trees in different conditions and minimise the damage of tree root intrusion after further multifactor investigations.

1. Introduction

The water network starts from the water sources, including groundwater systems (bores and wells) or surface water systems (rivers and reservoirs), via water mains through water treatment plants and water pumping stations to the end users/customers in both residential and industrial areas [1]. Sewage (wastewater) contains harmful microorganisms such as bacteria and viruses. Sewer pipe networks bring the wastewater from households to the wastewater treatment plant before discharging it into the environment. In sewer systems, pipe failures caused by tree root intrusions are the most common issue [2]. Direct contact with sewage overflow can result in serious public health issues which can be a concern to property owners and water utilities. In addition, tree root intrusion within a property causes extra expenditure for both water utilities and residents to repair the pipes and maintain the service.

In typically established and well-designed Australian urbanised residential areas (detachable single-family homes, each set back from the road with gardens and lawns), trees and shrubs are part of the urban landscape, and in general, they are required not to be planted closer than one metre to any water infrastructures including water mains and connections, according to the tree planting guidelines published by an Australian State water utility [3]. Large trees have large structural roots which can lead to the displacement of near-surface pipes leading to pipe breakage or leaks [4]. Thus, urban design professionals must understand the basic needs of trees and principles of planting to receive the benefits from trees without unduly risking infrastructure damage.

Today, remediating problems caused by tree root intrusion is costly. In Californian cities, $1.64 per capita was spent to solve the problems caused by tree roots in 1996 [5]. In Sweden in 2003, EUR 5.6 million was spent on problems caused by tree root intrusions [6]. EUR 28.4 million in 2009 was spent in Germany to remove the tree roots in pipe systems and maintain the pipes [7]. In Adelaide, the capital of South Australia, more than 30,000 cases of tree root intrusions were reported by the local water utility, which accounted for 58.9% of all pipe failures dealt with by the South Australian Water Corporation (SA Water) in 2020 [8]. SA Water spent $2.3 million in 2020 to clean up roots in the pipe network, $2.25 million per year on protective maintenance, and $200 k annually for compensation to customers for issues related to tree roots [8].

The aim of this review is to investigate the feasibility of predictive analytics that can be utilised to improve the solutions of tree root intrusion for the water utility industry. Establishing the predictive relationship between tree root intrusions and related factors could assist urban design professionals in the risk evaluation of tree root intrusions under certain circumstances and in decision-making. An overview of tree root intrusion and its maintenance strategy options are presented to identify the influencing factors and some published empirical models that predict the general risk to pipe networks without the consideration of trees.

2. Methodology

This review performed a literature search for the keywords, including “tree root intrusion”, “root damage”, “pipe system”, “sewer pipe”, “maintenance”, “influencing factor”, and “predictive model” (Figure 1), in databases including Google Scholar, Web of Science, ScienceDirect, and ResearchGate. The literature search was performed through the hand-searching of potentially relevant references. In terms of screening, any type of literature (journal articles, technical reports, and engineering records) was eligible as long as the focus was on tree root damage (mainly of street trees) in urban environments and its maintenance strategies. This review included 80 articles and 2 reports covering a time span of the past 47 years. No linguistic restrictions were imposed. This review included materials in English, Chinese, and German.

The current worldwide remediation options to manage tree root intrusion can be further summarised into two categories: reactive and proactive solutions. Reactive solutions include mechanical or chemical techniques to clean the pipe systems and control root growth after tree root intrusion damage is reported. Proactive solutions include preventive approaches in urban design and asset management.

This study focuses on proactive solutions to reduce the risk of tree root intrusion through predictive modelling. Thus, an overview of the potential factors is presented from previous studies that have identified and investigated the effects of various factors on tree root intrusions. The number of tree root intrusions is differentiated based on the change in these factors. The factors were categorised into three groups: (1) tree factors, (2) pipe factors, and (3) environmental factors. Research and analytical tools from previous studies that can be used to process the data and establish the relationships were introduced.

3. Results

3.1. Literature Explored

An overview of the literature cited in this review related to the main issues is presented in

Table 1. Summary of the literature relating to key issues.

Findings		Literature
Overview of issues related to tree root intrusion		DeSilva et al. [2]; Stål [6]; Rodríguez et al. [9]; Kuliczkowska & Parka [10]; Obradović [11]; Harris et al. [12]; Östberg et al. [13]; SA Water [14]; Schrock [15]; Myrans et al. [16]; Hammond et al. [17]; Stein [18]; Selvakumar et al. [19]; Ko and Standing [20]
Root Detection Technologies		Kuliczkowska & Parka [10]; Myrans et al. [16]; Zou et al. [21]; Lantini et al., [22]; Alani and Lantini [23]; Borden et al. [24]; Wu et al. [25]; Moore and Ryde [26]; Hirano et al. [27]; Randrup et al. [28]; Duran et al. [29,30]
Maintenance strategies	Reactive approaches	Kuliczkowska & Parka [10]; Obradović [11]; Selvakumar et al. [19]; Cameron et al. [31]; Santos et al. [32]; Syssner and Jonsson [33]; Rolf and Stål [34]; Randrup et al. [35]; Costello et al. [36]; Evans et al. [37]; Mitchell and Schnelle [38]; Widhiastuti et al. [39]; Ducoste et al. [40]; Groninger and Bohanek [41]; Ake [42]; Randall et al. [43]; Yeung and Criddle [44]
	Proactive approaches	Rolf and Stål [34]; Ward and Clatterbuck [45]; Svihra [46]; Torres et al. [47];
Factors influencing tree root intrusion	Root growth patterns	Trowbridge & Bassuk [48]; Fityus et al. [49] Miller and Neely [50]; Gilman [51]; Kozlowski [52]; Gao et al. [53]; Shi et al. [54]; Fransson [55]; Ni et al. [56]; Wang et al. [57]; Rogers et al. [58]; Taylor et al. [59]; Crookshanks and Taylor [60]; Crookshanks et al. [61]
	Tree species	Stål, [6]; Östberg et al. [13]; Randrup et al. [35]; Fityus [49]; Ridgers et al. [62]; Mattheck and Bethge [63]; Sjöman et al. [64]; Baker [65]
	Pipe material	Östberg et al., [13]; Randrup et al. [28]; Rolf and Stål [34]; Pohls et al. [66]; Bradshaw et al. [67]; William [68]; Marlow et al. [69]
	Pipe age	Rolf and Stål [34]; Pohls et al. [66]; Marlow et al. [69]; Gould [70]; Weerasinghe et al. [71]
	Pipe diameter	Rolf and Stål [34]; Pohls et al. [66]; Marlow et al. [69]
	Tree-to-pipe distance	Kuliczkowska and Parka [10]; Torres et al. [47]; Pohls et al. [66]
	Soil profile	Kuliczkowska and Parka [10]; Baker [65]; Pohls et al. [43]; William [68]; Lytton [72]; Watt et al. [73]; Wang and Smith [74]
	Weather conditions	Gould [47]; Nowak [54]; Baker [65]; Gould [70]; Kargar and Joksimovic [75]; Pregitzer et al. [76]; Xing et al. [77]; Ossola et al. [78]; Moore [79]; Streckenbach [80]; Nowak [81]
	Other environmental factors	Obradović [11]; William [68]
Predictive modelling using pipe factors		Tran et al. [82]; Gedam et al. [83]; Robles-Velasco et al. [84]; Tran [85]; Malek Mohammadi et al. [86]; Agresti [87]; Hosmer et al. [88]

.

3.2. An Overview of Issues Related to Tree Root Intrusion

Tree root intrusion is identified as the main reason for sewer pipe blockages [2] and managing blockages and structural failures has become a frequent problem in the maintenance of sewer systems and their functionality [9]. It has been found that sanitary sewers have the highest risk level of malfunctioning while stormwater sewers have the least risk [10]. Sewer chokes, involving tree roots penetrating and invading sewer pipes, have caused sewer problems including sewer stoppages and overflows, structural damage caused by growing roots, the formation of septic pools behind root masses, a reduction in hydraulic capacity, and the loss of self-scouring velocities [11].

Roots proliferate in areas suitable for growth [12]. Wastewater contains necessary nutrients for trees to grow, with sewer pipes having the potential to provide trees with those nutrients. The condensation of moisture on the surface of pipes and in the soil near the pipe systems can attract tree roots. If there is a crack in the pipe, tree roots can easily find their way into the pipe searching for nutrients and consequently damage the pipe system (Figure 2 as an example). In addition, tree root intrusion in pipe systems is more likely to occur in damaged/leaking pipes or those with faulty joints [13].

The Water Research Centre in the United Kingdom has researched trends up to the mid-1980s and recorded approximately 5000 pipe collapses and 200,000 blockages, with a 3% annual increase [15]. Tree roots, pipe corrosion, soil movement, and inadequate construction combined to cause the most structural failures. More than 40% of sewer pipes experienced damage caused by tree roots and 50% of all stoppages in sewer systems in the United Kingdom were related to roots [15,16].

Several previous research studies have also recognised that tree root intrusion can potentially lead to the flooding and pollution of neighbourhoods [6,17]. Wastewater exfiltration from damaged sewer pipes due to partial or total blockage of flow has the potential to contaminate the surrounding soil and groundwater [18,19]. Root intrusion can expand any existing fissures in pipes and joints in pipe systems, leading to surrounding soil entering the system through the defect. The construction of pipelines and the backfilling of trenches can leave pore spaces in the soil, allowing tree roots to reach pipes. Roots can still damage the pipe structure and result in soil erosion in the vicinity of sewer pipes, consequently leading to structure breakage and collapse [15].

Interactions between roots and soils can also pose a threat to pipe structure. Long-term interaction may involve fluctuations in soil moisture, causing the shrinkage or swelling of soil [20]. A pipe becomes stressed and easier to break when the surrounding soil shrinks and swells. Tree roots will find pathways after the soil shrinks and swells, especially in dry soils that have settled due to shrinkage, and the roots extend and grow larger [20].

Construction quality is also important. Inefficient construction could lead to soil disturbance, creating pathways for woody roots to grow towards pipes. In addition, poor construction practices can result in inappropriate connections and leakage from joints, causing moist soil profiles and allowing roots to grow into a pipe structure [6,15].

3.3. Root Detection Technologies

Destructive methods can provide direct and accurate mapping of the root systems [21], which involves excavation and uprooting [22,23]. However, these techniques are onerous and time-consuming [23]. Digging and trenching can affect the surrounding trees and may cause irreversible damage to trees [21].

On the other hand, non-destructive testing methods, such as magnetic resonance imagery, X-ray computed tomography, and ground penetrating radar, have been successively used to detect root systems without harming the tree [24,25]. Among these, ground penetrating radar (GPR) is a fast, reliable and cost-effective technology [21,22], widely used in locating buried objects such as bedrock, subsurface water levels, and utilities infrastructure and objects [23]. It can identify roots with a smallest diameter of 0.5 cm [26]. However, factors of root diameter, root water content, and intervals between roots can lead to an underestimation of root biomass when using this external detection tool [27].

Another currently used technology in sewer inspection is closed-circuit television (CCTV) [10,16,28]. The operators of CCTV will inspect the pipes offline by using a small camera that travels across the pipe, recording the images on a videotape [16,29]. Alternatively, operators can drive a remote-controlled pipe inspection gadget (PIG) through the network [16]. PIGs can accurately measure the internal dimensions of a pipe by equipping laser profilers [30] and record 360-degree panoramic footage of pipe interiors with 3D cameras [16]. This method can accurately detect tree roots in pipe systems but is time-consuming, expensive, and prone to human error [16,29].

3.4. Maintenance Strategies

The current maintenance strategies can be summarised into two general maintenance strategies: reactive and proactive approaches [31]. Reactive approaches include mechanical techniques to remove the root masses in the pipe systems and enhance the pipe and joint resistance to tree root intrusions. In addition, chemicals may be applied to clear the roots and suppress root growth. A proactive approach is a preventative strategy performed in the landscaping and early construction stage to minimise the damage of tree root intrusions in the future.

3.4.1. Reactive Approaches

(i)

Mechanical Techniques

Mechanical techniques are the most commonly used method to clear root intrusions into pipes [32,33]. Roots are cut and removed from the inside of the structure by implementing tools and devices such as drilling machines, rodding machines, jetters, and winches [11]. This method is usually applied as the initial action when the structure requires immediate action and remains open for service. However, Rolf and Stål [34] pointed out that root cutting is only a temporary solution as tree roots become stimulated and form new and more fine roots after old ones have been cut. So, the situation can worsen and lead to high maintenance costs with the need for repeated tree root removal.

Several modern remediation alternatives were discussed and reported by Rolf and Stål [34]. Pipe relining is a less intensive method than traditional pipe repair methods and reduces future intrusions. A resin-saturated felt tube made from a suitable material (generally polyester or fibreglass) for resin impregnation is inserted into a damaged pipe. This resin then hardens and forms another pipe within the damaged pipe. This method can minimise the risk of further tree root intrusion; however, the relining material is highly expensive. Another method is called pipe slip lining. This method utilises plastic pipes that are either pushed or drawn through the existing damaged pipelines. The pipelines can be easily installed, and the required tools are widely available. However, with both techniques, the new pipes will generally have a reduced pipe diameter and pipe wall thickness.

Rolf and Stål [34] also suggested that a liquid dual-component acrylic gel called Penetryn can be utilised to seal the cracks in pipes and the gel can seal the leakage of the structure [19]. However, it cannot be used when tree roots have already penetrated the pipe. Lastly, pipe replacement is always an effective method when small sections of pipes have been damaged [34].

Randrup et al. [35] suggested the application of root barriers, which is a technique that is designed to prevent or delay the impact of roots on infrastructure to avoid tree root intrusion. Costello et al. [36] indicated that these barriers are constructed in the tree planting hole to minimise the horizontal expansion of roots and guide roots to grow downward once roots reach the edge of the planting hole and encounter the barrier. By implementing root barriers, the number of roots found outside the barrier is reduced by 35–55%. However, in a study on the vertical moisture barrier designed to solve a problem caused by expansive soil, the effectiveness of a barrier to resist tree roots was questioned. Tree roots were attracted to the barrier and encouraged to grow in the vicinity of the barrier in search of moisture [37]. The tree roots were able to seek out points of weakness in the membrane, a component of the deep barrier system. Small holes in the membrane or breaks in the welded joints were enough for the roots to penetrate the barrier.

(ii)

Chemical Application

Apart from mechanical techniques, the chemical technique is primarily based on herbicides and has advanced to the point where either applying both mechanical controls or chemical means alone can remove the root mass and control future growth [38]. Some of the most used chemical herbicides are Copper Sulphate, Diquat, Dichlobenil, Metam Sodium, and miscellaneous chemicals [10,39,40,41]. These chemicals are flushed into the pipe system and contact any tree root masses. Copper Sulphate will control roots intruded in a pipe over time with low water volume without harming tree health [38]. Diquat and Dichlobenil are two aquatic herbicides that are mainly used to control aquatic weeds and retard tree root growth in contact for long periods [41]. Metam Sodium can be implemented in many circumstances, but it can pose a hazard to the environment [10]. Other miscellaneous chemicals can also be applied to control roots. These chemicals include strong acid products, which destroy the tissues and cells of plants [39]. These strong chemicals can destroy roots in a rather short period of time but also provide only a temporary impact because these chemicals degrade rapidly [10].

Chemical controls can be effective, but these chemicals can be hazardous to the surrounding environment and may not be desirable in wastewater treatment systems, especially in nitrification and denitrification [39,42,43,44]. Different concentrations of chemical compounds and metal ions in herbicides can inhibit the activity of microorganisms in active sludge, disrupting their functions and impairing wastewater treatment processes [39,42]. In addition, many countries have already published strict regulations on chemical dosage. In Denmark, herbicide use on all public land was banned after 2003 [35]. The chemical control of roots within pipes is not allowed in Sweden [34].

3.4.2. Proactive Approaches

In landscape design, it is important to select tree species that pose less risk of intruding into a pipe. Planning and management in construction can prevent or minimise pipe blockages, tree removals, and pipe maintenance at a very early stage. Rolf and Stål [34] recommended that trees should be planted as far from underground pipes as possible. They also suggested that pipes should be constructed in a pattern that least interferes with trees and structures [34]. Regulated tree replacement can also be beneficial to control root chokes [45]. Ward and Clatterbuck [45] suggested that it is reasonable to replace and re-plant faster-growing species every 8 to 10 years.

When constructing a pipeline close to existing trees, Rolf and Stål [34] suggested that it is important to consider tree removal, as it is better to remove the tree and replace/replant with a new one, from an economic perspective, if the tree could be affected by the construction and may not be able to survive.

Appropriate pipe construction can reduce the risk of structure failure. Svihra [46] indicated that the joints of older pipes using rubber flanges and O rings are a major problem as they can open up with soil movement. Polyvinyl chloride (PVC) pipes in sewer systems solved the issue because of the highly flexible nature of PVC pipes, and their glued joints can reduce the likelihood of cracking after soil settling [46]. The careful placement and backfilling of pipes during construction can reduce the spaces in the surrounding soil leading to less risk of cracked pipes, thus increasing root intrusion resistance [47].

3.5. Discussion of Tree Root Intrusion Approaches

The existing reactive response can be summarised as the strategies with mechanical or chemical applications to clean the pipe systems and slow down the next tree root intrusion incidents, either by strengthening the pipe systems or by suppressing root growth. The mechanical techniques repair the pipe to the pre-damaged condition but require the damaged pipes to remain open for construction. Chemical applications clean up the root masses in the pipes but may leave the pipes damaged and subsequently pose a potential threat to the surrounding environment. These reactive strategies efficiently solve individual tree root intrusion incidents. However, these approaches can cause intrusions to re-occur in the long term, causing high maintenance costs or tree removals in the worst-case scenarios.

Preventative strategies select better tree species in urban design and choose appropriate pipe material and construction procedures in asset management. These strategies implemented at the planning stage can reduce or prevent the occurrence of tree root intrusion incidents in the long term and reduce repeated high maintenance costs.

3.6. Factors Influencing Tree Root Intrusion

This literature review aimed to summarise and establish a relationship between tree root intrusion and the various factors involved to provide improved knowledge of the problem. The main factors influencing tree root intrusions are summarised in

Table 2. Summary of tree root intrusion factors.

Tree Factors	Pipe Factors	Environmental Factors
Root growth patterns	Pipe material	Soil profile
Tree species	Pipe age	Weather condition
	Pipe diameter	Other environmental factors

.

3.6.1. Tree Factors

(i)

Root Growth Patterns

Understanding tree requirements and root growth patterns can be helpful in recognising the cause of tree root invasion and potential control methods. Six basic requirements for tree growth are identified by Trowbridge and Bassuk [48]: oxygen, carbon dioxide, light, water, nutrients, and appropriate temperatures. In many situations, tree root systems may be concentrated in the top metre of the soil profile if there is an abundant source of moisture; however, “the roots of trees and shrubs are not limited to particular distances or depths, but rather, they will extend as far as they need to, or as far as they can, depending upon the type and species” [49]. Indeed, researchers have found that the roots of a woody plant can grow two to three times its height laterally [50]. In some extreme cases, the root system of a tree can sometimes grow up to seven times the tree’s height [51]. This is dependent on tree species, individual tree characteristics, soil characteristics, and climate [52].

Plants can react to these environmental factors through a variety of positive and negative directional responses they have evolved [53]. The hydrotropism of root systems, which refers to the directed growth of roots in relation to a gradient in moisture, may play an important role in influencing the susceptibility of stormwater or sewerage pipes to tree root damage [54,55]. Specifically, roots of Schefflera heptaphylla under intense water stress at dense planting spacings, or during long drying periods, are more sensitive to hydrotropism [56], which may cause roots to preferentially grow toward pipes and other underground structures that provide water.

Plants can have the ability to adapt to environmental changes and avoid hazards [53], which represents a critical factor in root intrusions. Plants can respond to atmospheric concentrations of carbon dioxide, which is one of the basic tree root growth needs [57]. Previous studies have proved that tree root growth can be stimulated by increased concentrations of carbon dioxide [58]. The biomass (or root length per unit volume of soil) might also increase under the elevated carbon dioxide condition. Recent research has also shown that the number of lateral root growth of a one-year-old 20 cm hardwood cuttings of the Populus deltoides X nigra (P. euramericanu) was increased under high concentrations of carbon dioxide [59,60].

Crookshanks et al. [61] conducted research aiming to evaluate the relationship between carbon dioxide and tree root growth. They investigated tree root growth and the function of the three most planted species in Britain, oak (Quercus petraea L.), Scot’s pine (Pinus sylvestris L.), and ash (Fraxinus excelsior L.), in both ambient and elevated carbon dioxide concentrations [61]. After the measurements were analysed over several time periods, they found that the root length became greater in an elevated carbon dioxide concentration for all species. Root production, in the first 6 months, increased significantly for all three species. Although the magnitude differed among all three species, it was greatest in ash (Fraxinus excelsior L.). In 20 months, the roots appeared to have an increased production when exposed to elevated carbon dioxide, but the result in different species varied. For example, a large increase in coarse root mass was observed in Scot pine (Pinus sylvestris L.), but few fine roots were detected.

(ii)

Tree Species

Tree species form different tree root systems, which depend upon the environmental conditions—most importantly, the availability of water [49]. Östberg et al. [13] conducted a risk analysis on different tree species over a total length of 33.7 km in Malmö and Skövde, two cities in Sweden. The researchers identified the trees causing the intrusion excluding vegetation more than 20 m away from the intrusion. Pipes built from 1970 onwards were investigated as an improved pipe jointing system had been introduced then.

After analysis of the data, a total of 52 species of woody plants were identified as capable of intruding pipe systems. Östberg et al. [13] found that Malus floribunda Van Houtte and Populus canadensis “Robusta” Moench had the highest risk of intruding the pipe system, with the mean number of root intrusions per joint being 0.694 and 0.456, respectively. The risk was assessed by dividing the sum of all root intrusions by the number of all available pipe joints within a 20 m radius of the tree for that particular species. Thuja plicata D. Don had the lowest risk of root intrusion (0.065) and Crataegus punctate Jacq. had the second lowest (0.091).

Due to the result that all woody plants in the study area were capable of intruding pipe systems, Östberg et al. [13] suggested that no woody plants should be planted in some extremely sensitive areas where pipelines are nearby. The careful selection of tree species, therefore, was suggested instead to lower the risk of root intrusion [13].

Several researchers have investigated the ability of tree species, varying from fast-growing to slow-growing species, to invade pipe networks. However, contradictory results have been obtained in some situations. Researchers [35,62] indicated that Salix and Populus, both fast-growing species, presented a higher risk of intrusion than other genera. Stål [6] indicated that slow-growing and moderate-growing species, such as Tilia cordata, do not cause root intrusions. However, the survey conducted by Östberg et al. [13] of 26 Tilia cordata specimens, planted within a distance of 20 m from a pipe, generated a mean number of 0.365 root intrusions per available joint, which means that this species cannot be considered a low-risk species.

Malus was listed by Mattheck and Bethge [63] as the species that caused the least root intrusions in all woody plants. This observation was supported in part by Östberg et al. [13], as seen in the summary of their data for the species in

Table 3. Mean number of root intrusions of Malus species (adapted from Östberg et al. [13]).

Species	Number of Trees Within 20 m from Pipe	Mean Number of Root Intrusion per Available Joint
Malus domestica	55	0.189
Malus floribunda	11	0.694
Malus sargentii	5	0.195
Malus spp.	27	0.192

. A large number of Malus domestica were planted near pipe systems, but the mean number of root intrusions indicated that this species has a low risk of root intrusion (19%). However, according to this table, Malus floribunda was identified as a species with a high risk of root intrusion (69%). This shows that the risk of root intrusion between species of the same genus can be significantly different.

The reason for the contradictory results on fast-growing species and slow-growing species can be explained perhaps by the planting frequency of a certain species. Preferences for the selection of urban tree species can differ greatly between countries and cities. Sjöman et al. [64] indicated that both Salix spp. and Populus spp. were popular species before the 1970s in Malmö. These species accounted for 3.9% and 6.1%, respectively, of the total tree stock. Another reason can be the inconsistency of other variables that might impact tree root intrusions; for example, different weather conditions and soil characteristics.

These findings may not be readily applicable to other cities around the world, as the species having the highest risk in the previous discussion may not be as commonly planted. In a study in Adelaide, Australia, Baker [65] examined root balls and identified microscopically the likely offending tree or trees for each root intrusion. Salix Babylonica, Eucalyptus camaldulensis, Populus nigra italica and Athel pine were deemed to be the most aggressive, as they caused the highest proportion of intrusions at a distance greater than 7.5 m away.

3.6.2. Pipe Factors

(i)

Pipe Material

The risk of tree roots intruding into a pipe depends on several root and soil characteristics. However, the ability to resist the intrusion varies according to pipe and joint characteristics as well. According to research conducted by [28], pipe materials are significantly different. Previous research has found that vitreous clay pipes and concrete pipes without rubber seals have relatively more blockages [34,66,67].

Before 1960, worldwide sewer constructions involved mainly vitrified clay, brick, and concrete [34]. Thereafter, plastic, ductile iron, steel, and reinforced concrete were implemented in sewer construction [34]. Early technology, design, and the availability of raw materials limited the choice of pipes, which consequently resulted in the inflexibility of the old pipes to withstand deterioration in these old systems [68]. The pipe materials used in old cities do not have the capability to tolerate the force produced by tree roots. These old pipes moved waste, but they were not completely watertight in the pipe joints.

An exploratory investigation conducted by Marlow et al. [69] in Australia further supported these views. Pipe blockages were investigated for five sewer pipe materials: concrete, vitreous clay (including earthenware and salt-glazed ware materials), asbestos cement, PVC, and polyethylene. It was shown that both vitreous clay and concrete pipes had significantly higher numbers of blockages and blockage rates compared to PVC and polyethylene pipes.

Östberg et al. [13] examined the risk of intrusion for two materials, concrete and PVC. Pipe age was also considered. The study concluded that there was a significant difference between PVC and concrete pipes. PVC pipes had a significantly higher number of intrusions into joints (mean number of 0.661 root intrusions per joint) than the joints of concrete pipes experienced (0.080 root intrusions per joint). However, this finding was skewed in that PVC pipes had a much higher number of service connections at quite shallow depths resulting in a higher proportion of PVC pipes susceptible to woody plants.

(ii)

Pipe Age

Pipe age significantly affects pipe failures, especially for cast iron and asbestos cement pipes [70]. Most old pipes are made of these two materials. The possibility of these materials deteriorating over time is relatively higher than other pipe materials. Therefore, these old pipes have a higher possibility of failure as the pipe age is a major factor affecting the strength of the pipe [71].

Pohls et al. [66] identified the percentage of blockages of sewer systems in Melbourne, Australia, based on the age groups and total percentage in each age group. The findings are summarised in

Table 4. Blockage rates for pipes in different age groups (adapted from Pohls et al. [66]).

Age Group (Years)	Percentage of Pipes in the Age Group (%)	Blockage (%)
0–9	18.3	0.9
10–19	21.8	10.3
20–29	18.9	24.2
30–59	20.1	49.3
60+	20.9	15.3

. Each age group had a similar proportion of pipe lengths (18.3 to 21.8% of the overall total length). A total of 125 tree root intrusion sites were visited; 117 of the sites had trees that may have caused the blockages and the remaining 8 were excluded due to an inability to access them or there being no trees present.

There is a trend of an increase in pipe blockages in age groups as the numbers of years increase to the last group (30 to 59 years), where the peak percentage blockage was 49.3%. Based on this result, it was suggested that the more recent the construction of sewers, the fewer the blockages [66]. In addition, the chance of a pipe being intruded in the first 20 years of service is relatively low because mature trees are less likely within newly developed assets and sewer flows are relatively manageable [69].

Rolf and Stål [34] did similar research in Malmo, Sweden. In this study, three classes of intrusion were defined based on the size of roots that had intruded, as follows:

• Class 1: small and few roots in the pipe but no water leakage.
• Class 2: coarse roots penetrate further into pipes and are in the water flow.
• Class 3: large roots or numerous roots at one site.

The results summarised in

Table 5. Pipe intrusion against pipe construction year (adapted from Rolf and Stål [34]).

Pipe Construction Year (Age in Year)	Number of Intrusions per 1000 m
	Class 1 Intrusion	Class 2 Intrusion	Class 3 Intrusion
1910–1919 (70 yrs.)	11	2	1
1920–1929 (60 yrs.)	10	1	0
1930–1939 (50 yrs.)	6	1	1
1940–1949 (40 yrs.)	11	2	1
1950–1959 (30 yrs.)	8	2	1
1960–1969 (20 yrs.)	8	2	1
1970–1979 (10 yrs.)	1	1	0

showed that a peak number of Class 1 intrusions appeared in pipes constructed in 1949 and earlier, although pipes constructed between 1930 and 1939 fared relatively better. The number of intrusions was generally less in pipes constructed between 1970 and 1979. In short, the research concluded that there appeared to be more cases of intrusions in older pipes.

(iii)

Pipe Diameter

Pipe diameter was recognised as one of the factors that is correlated with pipe blockages. Marlow et al. identified that pipes 100 mm in diameter have the highest failure rate, and the failure rate reduced as the diameter increased [69]. Pohls et al. [66] indicated that pipes 150 mm in diameter experienced 88.5% of blockages while accounting for 42.3% of all pipe diameters.

Rolf and Stål [34] investigated the relationship between the number of root intrusions and pipe diameter as shown in

Table 6. Pipe intrusion against pipe diameter (adapted from Rolf and Stål [34]).

Pipe Diameter (mm)	Number of Intrusions per 1000 m
	Class 1 Intrusion	Class 2 Intrusion	Class 3 Intrusion
225	11	2	1
300	8	2	1
375	11	3	1
400	7	4	1
450	6	1	0
500	5	1	1
600	7	1	1
750	2	1	1

. Pipe diameters ranged between 225 mm and 750 mm. The trend was for a decrease in the number of intrusions with an increase in pipe diameter. Root intrusions are more likely to occur in smaller-dimension pipes than in larger pipes. A possible explanation is that larger pipes are usually used as primary sewer lines or trunk sewers deeper in the soil and are difficult for tree roots to reach [34].

Large root penetrations (Class 3) were rare and were equally distributed in almost all types of pipes with just one intrusion being observed for each diameter. The exception was for 450 mm diameter pipes (no Class 3 intrusions).

(iv)

Tree to Pipe Distance

Kuliczkowska and Parka [10] found that the number of intrusions decreased when the horizontal distance between the sewer pipe and the tree increased. Torres et al. [47] stated that bigger trees require longer distances to reduce the risk of tree root intrusions because bigger trees can extend more laterally to search for water and nutrients. Pohls et al. [66] supported the decreasing trend between the number of blockages and increasing distance. A total of 125 tree root blockage sites were visited. Of the sites examined, 117 showed potential blockages caused by nearby trees. The remaining eight sites were excluded from the analysis due to no access or a lack of tree presence in the vicinity. The percentages of recorded pipe stoppages are shown against distance ranges in

Table 7. Percentage of root intrusions at the distance between tree and stoppage (adapted from Pohls et al. [66]).

Distance Between Tree and Stoppage (m)	Percentage of Recorded Pipe Stoppages (%)
0–2	7
2–4	37
4–6	23
6–8	11
8–10	7
10–12	8
12–16	4
16–24	2
24–36	1

[66].

3.6.3. Environmental Factors

(i)

Soil Profile

Several requirements for tree root growth have been identified: nutrients, soil structure, weather conditions, and artificial conditions [68]. Tree root systems are important structures stabilising the tree body and transporting necessary nutrients from the soil. Trees will obtain carbon, hydrogen, and oxygen above the ground while their roots will mine the soil searching for water and nutrients below the ground. Lytton [72] suggested that the interaction between a soil profile and root system influences the growth pattern of vegetation. A proper proportion of soil structure, which is combined with soil particles (varying from clay-like particles to sand-like particles), organic matter, air, and water, can form a medium that is conducive to tree root growth. Adequate soil depth also provides a comfortable environment for tree roots to grow [72].

One of the most important soil characteristics influencing the propensity for sewer blockages is soil strength [73,74]. This same soil characteristic, as represented by the level of soil compaction or density, has been proven to impact the potential for tree root growth [73,74]. Theoretically, tree root growth tends to be relatively more trouble-free in less compacted soil as it can provide more space and oxygen for roots to penetrate. Baker [65] investigated the choke rates of seven prominent street tree species planted in two major soil types in the City of Adelaide, South Australia: sandy clay and heavy (highly plastic) clay. Sandy clay soil has a coarser texture and should be less compacted than heavy clay. Baker [65] found that only two of the species (Agonis flexuosa and Jacaranda mimosifolia) presented a higher number of chokes and choke rates in sandy clay soil, while the remaining five species (Celtis australis/occidentalis, Eucalyptus spathulate, Fraxinus oxycarpa/“Raywood”, Melia azedarach, and Platanus “Hybrida”) presented higher numbers of chokes and choke rates in heavy clay soil. These findings suggested that tree roots in compact soil extend towards the less restricting trench backfill and proliferate near the pipe system, therefore increasing the probability of tree root intrusion. In addition, soil shrinkage in heavy clay due to moisture extraction caused by large trees may cause the displacement and fracturing of a pipe, which increases the probability of tree root intrusions.

Kuliczkowska and Parka [10] summarised the factors of soil that contribute to tree root intrusions in sewer pipes in Poland. They found that silty soils and fine sandy soils accounted for half of the soil types in their study; those soil types have lower levels of compaction, and it was found that these soils experienced the largest number of tree root intrusions.

(ii)

Weather Conditions

The incidents of sewer blockages followed a seasonal pattern, with an increased number of incidents during the summer (wet and warm) months [75]. Summer storms can overwhelm the sewer infrastructures and flush large amounts of debris into sewer systems, leading to a seasonal rise in pipe blockages [75]. Another contributor is the tree root growth affected by elements of weather conditions such as temperature and precipitation. Tree root growth accelerated during the warmer months [76]. The fine roots of woody plants increased shallow root growth when precipitation increased [77]. The expanded volume of tree roots can increase pressure in sewer pipes, leading to more frequent blockages [78].

Seasonal soil environment changes can affect tree root growth. Baker [65] also observed that the maximum number of chokes caused by tree roots occurred in late autumn to early winter (wet seasons) and fewer chokes happened in late spring to early summer (dry seasons). Gould [70] indicated that pipe failure rates bottomed in wet seasons (May to December) when soil moisture is the highest and peaked in dry seasons (January to May) when the soil moisture is lowest. Enough moisture in the soil leads to horizontal root growth, while inadequate moisture availability results in vertical root growth [79,80].

Vertical root growth can be commonly found in urban areas because most of the soil surfaces are impervious. Pipe systems in urban areas reduce the availability of moisture in the soil by directing surface runoff away from trees. Leaves and other organic wastes are typically removed as well. Reduced nutrients and moisture availability in topsoil result in more vertical root growth in urban areas. These conditions are further exacerbated in Australia, as many regions have been subjected to prolonged drought [81], depriving trees of water for extended periods. Due to this reason, trees within the urban landscape are progressively becoming more competitive in finding water sources.

(iii)

Other Environmental Factors

Man-made obstructions can interfere with the ability of tree roots to obtain what the roots need, causing abnormal growth patterns [68]. Tree roots are drawn to the soil near sewer pipes because sufficient moisture and warmer temperatures can often be found [11]. If the pipe is cooler than the surrounding soil, the condensation of moisture happens, creating conducive conditions for tree roots. Fissures on the pipes can also provide nutrients and oxygen to invading tree roots. Tree roots will follow the pipe and penetrate the pipe if cracks or fissures are nearby. Subsequent root growth may widen fissures.

Unpredictable incidents, such as earthquakes, floods, and soil movements, can cause damage to sewer systems [68]. Inappropriate network design and the inadequate selection of pipe material can reduce the stability of the pipe system.

3.7. Integrating Pipe Knowledge into Condition and Failure Modelling

In the previous section, factors that can influence tree root intrusion were introduced. In order to establish a predictive relationship, some analytical tools are needed. Researchers have used statistical tools with databases to determine correlations between individual factors and analyses to establish any potential relationship between multiple variables. This section introduces three examples in the published literature. Unfortunately, neither consider factors associated with trees nor soil conditions.

3.7.1. Case Study, City of Greater Dandenong (CGD) in Victoria, Australia

Closed Circuit Television (CCTV) inspections between 1999 and 2007 of 417 wastewater pipe segments, equal to 3.4% of the total length of wastewater pipes in CGD, formed the database for this preliminary data analysis [82]. Correlation tests were performed to investigate the relationship between input variables as a preliminary data analysis. Linear relationships were sought among scale variables of the structural deterioration of pipes, including pipe size, pipe age, pipe depth, pipe slope, and tree count [82]. Tree count was the number of trees around the pipe blockage, which varied from 1 to 22. Pipe diameters ranged between 225 mm and 1950 mm. The pipe age ranged between 0 and 65 years. The pipe depth ranged from 0 to 4.83 m. Pipe slopes were horizontal or up to a maximum of 22.85%.

The test result indicated that statistically significant relationships were found between pipe size and depth, pipe size and age, pipe size and slope, pipe age and depth, and pipe depth and slope [82]. The study did not formulate a predictive equation but suggested that the relationships were reasonable because larger-size pipes, for instance, are often used as main pipes and buried deeper.

3.7.2. Case Study, India

Multiple linear regression analyses were based on data on the condition of sewer pipes obtained from Closed Circuit Television (CCTV) inspection [83]. The length of the inspection or its location was not specified. However, the sample size was 155. Pipe conditions were rated from one (excellent, with no defects) to five (bad, with damage threatening safety and functionality).

Pipe diameter, pipe age, and pipe installation depth were considered to develop the prediction model of pipe condition. The pipe material included concrete pipe and clay pipe. The pipe diameter ranged between 450 mm and 1800 mm. The installation depth varied from 65 cm to 213 cm.

The multiple linear regression analysis generated a model as shown in Equation (1):

Pipe Index = 0.678 + 0.092 × pipe age + 0.000095 × pipe depth

(1)

Interestingly, pipe diameter was found to be statistically insignificant and was excluded. Pipe age was the most significant variable. This model could explain 87% of observed pipe conditions based solely on pipe age and burial depth. The database and analysis did not include any aspect of vegetation or other environmental factors.

3.7.3. Case Study, Seville, Spain

A case study was performed to predict pipe failures in the city’s water supply systems [84]. Logistic regression and support vector classification were chosen because the output of these two methodologies can be interpreted as a probability ranging from −1 to 1, while 0 indicated no correlation. In addition, both predictive methodologies were suitable to deal with small and medium size databases and unbalanced data. In the case study, the predictive models were established for three materials (cement, metal, and plastics) and globally (entire networks) to investigate the failure probability in correlation to pipe age, pipe length, pipe diameter, and pipe pressure fluctuation. The water supply network, 3800 km, contains 4393 pipe failures. The pipe age ranged between 0 to 120 years old. Pipe length varied from 0 to 2500 m. Pipe diameter ranged between 0 to 2000 mm. Pipe pressure fluctuation ranged from 0 to 60 m (1 m = 9806.38 Pa).

The result of the study indicated that the models had excellent predictive abilities. The logistic regression model predicted 76.9% of the pipes and support vector classification predicted 75% of pipes. By applying these two models to the data in 2018, the logistic regression model prevented 34.09% of breakages by replacing 3.16% of the network’s pipes, and the support vector classification model prevented 29.52% of breakages by replacing 3.84% of the network’s pipes. The logistic regression model also indicated that the pipe material seemed to be the most significant variable followed by pipe length and pipe age. Pipes with smaller diameters are more likely to break. Pipe age is not relevant to cement or plastic pipes but is important for flanged end pipes.

3.7.4. Model Discussion

The advantage of linear regression lies in its simplicity, as it allows for an easy interpretation of the relationship between the dependent and independent variables. However, the simplicity of the model is the major disadvantage in defining the relationship between tree root intrusion and its influencing factors due to the complexity of the issue [85]. In addition, several tree root intrusion factors contain categorical data which the linear regression model presented a limited ability to classify and forecast [86]. The linear regression model is sensitive to outliers and its results may not describe a strong relationship for multiple input variables. Therefore, the linear regression model may not be a suitable model for describing and predicting the relationship between tree root intrusion and its factors.

Logistic regression is the most commonly used regression model for analysing datasets with two or more discrete outcome variables [87,88]. In logistic regression, the coefficients are directly related to the odds ratios, making it possible to identify the most influential factors in pipe deterioration [88]. This feature contributes to a more comprehensive analysis of sewer deterioration and supports the identification of essential data requirements for effective inspection and data collection [86]. The limitation of logistic regression is its reliance on large datasets to ensure model stability and generate meaningful results. Additionally, the linear structure of logistic regression limits its ability to capture complex or nonlinear relationships between variables.

4. Research Gaps and Future Study

While a wealth of research exists regarding tree root intrusion and its maintenance strategies, several research gaps still warrant further investigations.

The effects of tree root intrusion factors can be further analysed. Although numerous studies have been conducted on the analysis of the risk of tree root damage in different categories, significant uncertainty of the effects of factors remains due to the multifaced and complex nature of tree root intrusion. In real scenarios, tree root intrusion could be the result of factors in several categories altogether. There is a need to better understand the combined causes of tree root intrusion and the extent to which they contribute to its occurrence. This analysis can help to improve the risk assessment of tree root intrusion in different scenarios and assist in defining the weightings of tree root intrusion factors in future studies.

The tree root intrusion damage and its influencing factors in different regions can be further assessed and investigated. The preference for maintenance strategies may differ from region to region because factors like tree species and environmental factors varied. Different tree species have different root growth patterns, while pipe constructions may be dependent on factors like soil profile or weather conditions. By further assessing and investigating the regional differences, it is possible to categorise and rank the primary factors contributing to tree root damage by region. This analysis can assist professionals in local utilities to determine which factor has greater influence on tree root damage and, therefore, to select better maintenance strategies.

5. Conclusions

The aim of this review is to seek suitable solutions to tree root intrusions which continue to cause significant numbers of pipe blockages and failures, resulting in high costs of remediation and maintenance for governments and water utilities. This paper reviews and compares the existing maintenance strategies to seek out solutions to effectively manage tree root intrusion. Modern ways of mechanical controls involve root mass removal and techniques like pipe relining to prevent or delay the damage caused by tree roots. These controls provided a partial solution to this problem. Some countries and regions utilise herbicides together with mechanical controls to slow down root growth and prolong the service life of the pipe. However, these current reactive approaches do not solve the intrusion completely and leave the issue to re-occur. In addition to reactive approaches, proactive approaches of urban design and asset management were introduced. These preventative strategies minimise the damage of tree root intrusion by regulating the tree plantation in the design stage. Understanding that reactive approaches can only be a partial solution to this issue, this study decided to explore solutions from proactive approaches using predictive models to evaluate the risk of planting trees in different conditions.

Recent studies that investigated tree root intrusion have identified various factors that can influence and cause tree root intrusions in sewer systems. These factors were categorised into three major aspects: tree factors, pipe factors, and environmental factors. Correlations have been limited to considering pipe factors such as pipe diameter, pipe material, and pipe age. Previous studies have shown that logistic regression can be a more suitable tool than linear regression to formulate and predict the condition and failure of pipes. There are no examples found in the literature that include environmental or vegetation factors to predict the potential for tree root intrusion and pipe blockages.

Previous studies indicated that tree root intrusion can be influenced by many different factors. It may be possible to formulate models to predict pipe conditions and failures by incorporating these factors in the analyses. A predictive model would benefit local governments and water utilities by enabling the evaluation of the risk to pipes in different situations in the landscaping stage. With a sufficiently large database, it should also be possible to provide a more suitable selection of tree species in the future.