Maintenance management systems in evolution

1. 1Build the annual plan from quantities, not dollars. For each work activity, set a quantity standard (units of work per year per inventory unit) and an accepted average daily crew production rate based on a standard crew size. Then compute: (quantity standard × roadway inventory) / average daily production = crew days.
2. 2Express the annual work program as crew days per activity per geographical subunit. Allocate budget and resources to each subunit from those crew days, so a subunit with more inventory receives proportionally more resources.
3. 3Check your standards against the field before you adopt them. Performance standards in about 30 states had similar content but quite different optimal crew sizes, equipment mixes and expected productivity. Adjust quantity and performance standards for local facility conditions, traffic volumes, climate and drainage rather than copying another agency's figures.
4. 4Involve field supervisors in setting standards and in decisions on work methods, crew and equipment arrangements and work scheduling. Quality-circle techniques are one way to do this.
5. 5Authorize the plan in discrete units. Issue one crew day card per planned crew day for each activity (for example, 240 premix patching cards to a unit with 1,200 bituminous lane miles).
6. 6Schedule with a work calendar showing the crew days anticipated for each activity in each month of the year. Spend the cards against that calendar.
7. 7Control by comparison. Compare expenditures for each activity against budget at the district, division and statewide levels, with labor, equipment and materials reported separately.
8. 8Close the loop. Treat management as a cycle of plan, execute, control, evaluate, then replan. Use the control and evaluation results to refine the work quantities and the models that convert those quantities into budget dollars.

The Alabama Highway Department maintenance management system, put into operation in April 1973, works like this. Premix patching has a quantity standard of 1 ton per bituminous lane mile per year and an average crew production of 5 tons per day. A subunit with 1,200 bituminous lane miles needs (1 × 1,200) / 5 = 240 crew days. The plan is authorized by issuing 240 premix patching crew day cards to that unit. The work calendar spreads those 240 crew days across the months of the year. Reports then compare each activity's expenditures with its budget by district, division and statewide, with labor, equipment and materials shown separately. The source notes that the 5-ton rate depends on the assumed standard crew size. If your crew or equipment mix differs, the rate changes and the crew day count changes with it.

• If your crew size or equipment mix differs from the standard crew behind a published production rate, then recompute the crew days with a locally validated rate rather than reusing the borrowed one.
• If local conditions differ (traffic volume, climate, drainage, facility condition), then adjust the quantity standard for that region rather than applying one statewide figure.
• If a subunit's inventory is larger, then allocate crew days and resources in proportion to it, using the quantity standard times inventory.
• If actual expenditures on an activity diverge from budget in the control reports, then investigate the standard or the schedule and replan, rather than carrying the same plan into the next cycle.
• If you are considering contracting out work, then first use the system's data on staffing levels, equipment needs, workload distribution and work performance to estimate the effect on your own forces.

• Copying another agency's performance standards. The tell: the crew size and expected productivity differ from your actual crews, and planned crew days never match what the crews achieve.
• Planning in budget dollars with no work quantities behind them. The tell: you cannot say how much work a budget line buys.
• Setting standards without field input. The tell: supervisors ignore the work calendar and schedule by their own methods.
• Issuing a plan and never comparing results to it. The tell: expenditures per activity are not reviewed against budget, and the next plan repeats the same assumptions.
• Using one uniform standard across regions with different conditions. The tell: some regions chronically run short on crew days while others have slack.

• ▸Work activity: name the maintenance activity being planned.
• ▸Quantity standard: units of work per year per inventory unit (e.g., 1 ton premix per bituminous lane mile).
• ▸Roadway inventory: the count for this subunit (e.g., 1,200 bituminous lane miles).
• ▸Average daily crew production: the validated rate for your standard crew size and equipment mix.
• ▸Required crew days: (quantity standard × inventory) / daily production.
• ▸Cards issued: the number of crew day cards authorized, equal to the crew days.
• ▸Monthly calendar: crew days planned for each month, summing to the annual total.
• ▸Budget by labor, equipment and materials: the dollars allocated to this activity.
• ▸Actual versus budget (period): the expenditures, split by labor, equipment and materials, set against the budget.
• ▸Replan note: the adjustment to the standard, rate or calendar made after review.

1. 1Sort every road section into three bands before spending anything: sections with no observed fault (lowest priority, no action), sections with faults that drive rapid progressive failure such as blocked drains and potholes (immediate, fix on observation), and sections beyond maintenance treatment (reconstruction, kept out of the maintenance priority order but flagged).
2. 2Rank the remaining sections by priority. Observed condition data comes first, then the rule that available money is insufficient to treat everything. Draw the funding line (the 'money available' cut) down the ranked list and treat what falls below it as deferred, not forgotten.
3. 3Write each program as a performance budget line: program, objective, planned service level, and cost (the passage's example: '3,950 miles of lane line and pavement edge striping - $643,000'). Then attach resource needs by object of expenditure: people, equipment, materials, services, with last year's actual, current, and proposed values.
4. 4Allocate in two steps, following the Illinois MMIS design. Step 1: split the districtwide annual quantity for each activity across team sections, using each section's submitted needs and budget request plus its field needs survey of priority 1, 2, and 3 work. Step 2: have field engineers assign each section's quantity to specific roadway subsections, so the plan states how much and where.
5. 5Schedule when the work happens by spreading each activity's budgeted quantity across months using the team section's own four-year monthly history as a percentage model. For example, if history shows 11% of the work in October and 22% in November, a 90-square-yard budget gives about 10 and 20 square yards.
6. 6Check equipment and labor by quarter against the schedule. For each machine, compare hours required with hours available and read the over/short line (the Illinois report shows, for example, a mixer short by 13 in one quarter and a mower over by 25). Rebalance by moving work to quarters with slack or moving equipment between sections before the season begins.
7. 7Capture actual labor, equipment, and materials in one daily single-source report, convert them to dollars through automated resource cost tables, and compare work accomplished, productivity, unit cost, and total cost by activity against the plan. Re-run 'what if' scenarios when circumstances change instead of waiting for year-end.

Illinois DOT faced reduced revenue and rising labor, material, and equipment costs. Crews were cut, equipment replacement was postponed, and service levels were reduced to cover only the most urgent needs while the backlog grew. Its MMIS design addresses this with a repeatable allocation sequence. First, the districtwide annual quantity for each work activity is allocated to team sections, weighted by their submitted needs and by the quantities of priority 1, 2, and 3 work in a field needs survey taken just before allocation. Second, each section's engineers assign that quantity to actual roadway subsections on a microcomputer spreadsheet. Third, the spreadsheet spreads the work across months using four years of that section's history; with 11% of work historically in October and 22% in November, a 90-square-yard budget puts roughly 10 and 20 square yards in those months. Fourth, quarterly equipment reports show hours required against hours available, such as the concrete mixer short by 13 hours in one quarter and the tractor with mower over by 25. The manager can then see where equipment is overcommitted or idle and adjust the plan before the work is done.

• If a road section has no observed fault, then give it no action and the lowest priority; if the fault will cause rapid progressive failure (blocked drain, pothole), then treat it as top priority and fix it on or soon after observation.
• If a section needs reconstruction rather than maintenance treatment, then take it out of the maintenance priority order and flag it separately so it does not absorb routine maintenance funds.
• If funding cannot cover the full ranked list, then fund top-down to the money line and record what is deferred, rather than spreading money thinly across all sections.
• If the equipment report shows hours short in a quarter, then shift work to a quarter with slack or reallocate equipment before the season; if hours are over, then look for work to absorb the idle capacity.
• If conditions change (budget cut, weather, new priority 1 work), then rerun the work plan and resource needs with a 'what if' recalculation instead of editing individual lines by hand.

• Spreading effort evenly across all sections regardless of condition. The tell: sections with no observed fault receive work while urgent faults like potholes and blocked drains wait.
• Setting budgets with no performance link. The tell: the budget lists line items (labor, materials) but no service level or work quantity, so you cannot say what the money buys.
• Allocating annual quantities without a seasonal or equipment check. The tell: quarterly reports show large over/short hours on key machines, and crews wait for equipment or equipment sits idle.
• Using general goals with no measurable objective, as in the discussion's critique of a goal like reducing accidents through computerized inventory. The tell: no one can show whether the allocation changed condition or safety.
• Sizing tools and systems by price instead of defined use. The tell: a microcomputer bought at 256K memory later needs modification because uses were only vaguely defined, consuming money meant for maintenance.

• ▸Road section / location: identify the section or subsection being ranked.
• ▸Observed fault and band: record the fault and mark Immediate, Ranked, No Action, or Reconstruct.
• ▸Priority (1, 2, 3): assign from the field needs survey.
• ▸Activity and planned quantity: state the work activity, service level, and annual quantity (e.g., miles of striping, square yards of patching).
• ▸Unit cost and total cost: calculate quantity times unit cost to get funding required.
• ▸Cumulative cost vs. budget line: running total down the ranked list; mark where money runs out.
• ▸Resources required: list labor, equipment hours, and materials by quarter.
• ▸Hours available / over-short: compare equipment and labor available with required for each quarter and note the gap.
• ▸Monthly spread: apply the section's historical monthly percentages to the budgeted quantity.
• ▸Actual vs. planned review: record work done, productivity, and unit cost, and note adjustments.

1. 1Replace judgment-based needs with a measured condition baseline. Inspect the network, record surface distress, and load the results into a database that yields a condition index (PCI in PAVER). Distress is the prime condition indicator and public works inspectors can collect it, so the baseline stays maintainable in-house.
2. 2Set the minimum acceptable condition level before you price anything. The budget then answers a stated standard, not an open-ended wish list.
3. 3Build the budget from unit costs, not from last year's line items. Enter an average repair cost per square yard at each condition level for each surface type, taken from your own inspection reports, and run a multi-year projection (BUDPLAN produces a 5-year budget). Use it for routine, preventive and major repair costs alike.
4. 4Show the consequence of not funding. Produce the 'do nothing' network condition for major repairs and the 'continue historical' condition for routine and preventive work (FREQ report). State the damage likely to occur and the traffic restrictions or traffic management that unrepaired sections would force.
5. 5Show the effect of funding. PAVER cannot graph the effect of doing major repairs or raising maintenance directly, so build it manually from the predicted PCI in the ANALOC, CONLOC, CONDHIST and FREQ reports and plot it against the do-nothing curve (the passages' Figure 9 is an example).
6. 6Present dollars and work together. Put the line-item budget beside a performance budget: for each activity, the reason for the work, the planned service level, the quantity and frequency, the total cost and the line-item values. Accept the extra preparation time; this format is what shows the elected official the services the money buys.
7. 7Rank projects by stated criteria and explain the order. Group sections needing repair into logical projects, then show non-engineer decision makers that work follows identified need, not preference.
8. 8Audit the claim before submission. Test whether field-identified needs are overstated, whether quantities trace to inspection data, and whether the workload basis would survive an external challenge, because overstated needs are what destroyed Caltrans' credibility.

NTC Great Lakes loaded all its pavement data into PAVER during implementation. The initial inspection showed repair needs of $3.3 million for roads and streets. When funding became imminent, facility engineers and planners used PAVER to assemble one project for $800,000 in a matter of hours. The fiscal year 1984 budget was the first built this way: it rested on quantified projected needs and took a fraction of the usual time. The engineers justified the budget and projects with the current and projected network condition instead of the usual subjective, vague argument. They also showed commanding officers, who are not engineers, that priorities rested on identified need. That quantification was a prime factor in higher authority providing the funds to execute the 5-year road repair plan. Contrast Caltrans: its zero-based, quantity-standard system lost credibility because field forces overstated needs, and it fell back to a baseline-plus-increments approach that outside reviewers said lacked a supportable workload justification. In 1981 the legislature directed it toward a performance-based budget.

• If the budget request rests on field-reported needs with no measured condition data, then expect it to be challenged; collect inspection data first, because overstated needs sank Caltrans' system.
• If the audience is an elected official or a non-engineer commander, then show service level and consequence (condition with and without funding) next to the line-item total, not the line items alone.
• If funding is cut or a lump sum arrives suddenly, then draw projects from the prioritized plan using the database, as at NTC Great Lakes, where an $800,000 project was built in hours, instead of reopening the argument.
• If you use incremental (baseline-plus-increase) budgeting, then attach the work-load basis for the increment; an unsupported baseline invites outside scrutiny.
• If you need to show the benefit of an increase, then run BUDPLAN at several minimum PCI levels and plot condition with and without the spending; if the software cannot produce the graph directly, build it manually from the prediction reports.

• Quantity standards built on overstated field needs. Tell: outside reviewers or legislators question the workload documentation and the system is dropped, as with Caltrans' zero-based approach.
• Two-track budgeting. The financial budget goes to the elected official while the performance budget stays internal. Tell: decision makers see dollars with no link to services and cut across the board.
• Justifying with subjective, vague condition statements or standard fixes like a 2-inch overlay. Tell: the request cannot answer what happens if it is not funded.
• Showing only the do-nothing outcome and never the with-funding outcome. Tell: the audience sees the problem but not what the money changes, because PAVER does not graph the repair scenario directly.
• Unit costs not drawn from your own inspection reports, or data scattered across files. Tell: projections cannot be traced or reproduced, and engineers spend time searching records instead of analyzing alternatives.

• ▸Activity / project: name the work item and its network segment.
• ▸Reason for the work: state the condition or need driving it, citing the inspection date.
• ▸Current condition (e.g., PCI) and minimum acceptable condition: record both numbers.
• ▸Planned service level: state the target in measurable terms.
• ▸Quantity and frequency of work: give total units (e.g., square yards) and how often.
• ▸Unit cost basis: give the cost per unit and the source inspection report.
• ▸Total cost and line-item values: list the dollar total and the line-item breakdown.
• ▸Do-nothing / continue-historical outcome: state projected condition by year and the consequences, including traffic restrictions.
• ▸With-funding outcome: state projected condition after repairs, with the report used (ANALOC, CONLOC, CONDHIST, FREQ).
• ▸Priority rank and rationale: state the rank and the criteria that placed it there.

1. 1Scope the limits first. The supplied passages contain no liability doctrine; the Turner and Kramer paper on priority decisions in risk management for local governments is listed in the contents but its text is not included. Build the posture from the maintenance-management mechanics the passages do describe, and have counsel supply the legal standards separately.
2. 2Build a feature inventory of the assets that can fail (pavement, striping, signals, signs). The passages describe systems that update feature inventories automatically and tie work to location. Each asset needs an ID and a location so a defect report can be matched to a record.
3. 3Set a written service level for each activity, as a performance budget does: program, objective, planned service level, cost. A stated standard such as miles of striping per year gives you something to show you met or missed.
4. 4Capture work through single-source daily reporting. The crew reports labor, equipment, materials and accomplishment in one document each day. This produces the dated record of what was done, where, and by whom.
5. 5Use a work-order process that is created and monitored in the system, with priorities assigned by documented criteria. The passages cite programs that create and monitor work orders and help establish work priorities. Record why each item ranked where it did.
6. 6Reconcile maintenance records with the financial ledger so the two agree. Divergent systems commonly leave a 4 to 7 percent spread by function. If the numbers cannot be reconciled, your documentation of effort is open to challenge.
7. 7Before contracting out work, run the impact questions the passages list: effects on staffing and equipment, effects on work-load distribution, effects on peak and slack seasons, and whether the contractor will use standard work methods. Write down the answers so the standard defines the contracted work.
8. 8Review actual and projected conditions regularly with what-if runs on work plans, budgets and resource needs. Retain the runs that show a priority decision was made deliberately under stated resource limits.

The passages show a budget format that links objective, service level and cost. Program: 'Road and Street Maintenance.' Objective: 'Pavement line striping performed annually or as needed to delineate traffic lanes and thereby promote traffic safety.' Planned service level: '3,950 miles of lane line and pavement edge striping - $643,000.' That works out to about $163 per mile, so the stated standard has a price. The line item '101-320-002' lists a Mtce. Worker II at 1.2 resource quantity, with $19,872 actual, $20,050 current and $20,610 proposed. Used as a risk record, this tells you the safety purpose, the promised mileage and the funded resources in one place. If crews report striping completed daily by location, the agency can show what it committed to and what it delivered. If the funded level falls short of the stated standard, the gap is visible and can be documented as a priority decision rather than left unexamined.

• If the maintenance records and the accounting ledger disagree on spending (a 4 to 7 percent spread is common), then integrate them through single-source reporting and modified account codes before relying on either as evidence.
• If a service level is stated for an activity, then record actual accomplishment against it by location and date, because an unmet standard with no record is the weakest position.
• If you consider contracting out work, then answer the staffing, equipment, work-load and standard-method questions in writing before the decision, and attach the answers to the file.
• If resources fall short of the work program, then run what-if scenarios and record the resulting priority ranking and its rationale instead of letting deferral happen by default.
• If equipment utilization is low, then check whether the work program needs fewer units than are on hand before asserting an equipment need.

• Treating a budget as a risk record. The tell: line items exist but no stated objective, service level or location-level accomplishment connects them to a defect.
• Parallel maintenance and accounting books. The tell: managers and accountants cannot agree on dollars spent, and the spread is a few percent of total spending.
• Priorities with no stated basis. The tell: nobody can explain why one work order ranked above another.
• Relying on unreviewed standards, with effort or work-quantity standards estimated and never checked against pavement condition data. The tell: the planned service level has no data source.
• Contracting out work without a defined standard. The tell: no record of how the contractor's methods map to agency work standards.

• ▸Asset ID and location: the inventory identifier and where it sits.
• ▸Activity and objective: the maintenance activity and its stated purpose (e.g., striping to delineate lanes and promote safety).
• ▸Planned service level and cost: the committed quantity and funded dollars.
• ▸Condition or defect noted: what was observed, by whom, on what date.
• ▸Priority rank and basis: the rank given and the stated criteria behind it.
• ▸Work order number and date opened: the link to the monitored work order.
• ▸Resources assigned: labor, equipment and materials, from the single-source daily report.
• ▸Accomplishment and date closed: actual quantity done and when.
• ▸Deferral or shortfall rationale: if not done to standard, the what-if result or resource limit that explains it.

1. 1Define the output unit for each activity before you measure anything. Use the unit the work plan already uses: tons of asphalt placed, lane-miles mowed, or the quantity assigned to an activity such as 412 Full Depth Patch. Productivity is only comparable when the numerator is the same unit every time.
2. 2Define the input unit once and hold it fixed: man-hours, crew-days, or cost. Record all three if the system allows, because unit cost and productivity answer different questions. Do not mix crew-days in one period with man-hours in another.
3. 3Capture actual work accomplished and resources used by activity and general location, at the field level, on a daily or near-daily basis. The systems described in the source produce actual work accomplished, productivity, unit cost, and total cost as soon as the files are updated. Daily entry removes the need for a separate manual record.
4. 4Set the planned standard before comparing. Start from the inventory-times-quantity-standard calculation (maintainable features multiplied by average effort per feature). Then adjust it with condition survey data and local factors: facility condition, traffic volume, climate, and drainage. A productivity ratio measured against a standard that ignores condition tells you little.
5. 5Compute actual productivity (output ÷ input) per activity and per location, then compare it to the planned standard and to that activity's own history. Compare planned versus accomplished quantities, as the activity 412 sheet does by route segment (needs survey, budget, and accomplished columns by fiscal year).
6. 6Separate a productivity gap from a quantity gap. Check whether the crew placed less than planned (a work-plan adherence problem) or used more hours per unit than the standard (a productivity problem). The follow-up reports measure both plan adherence and performance standards.
7. 7Take the variances to the field supervisor. Ask about work methods, crew and equipment arrangement, and scheduling, which are the areas where the source says supervisors are directly involved and where some agencies use quality circles. Revise the standard if the field explanation holds, since standards are meant to be 'constantly evaluated and fine tuned.'
8. 8Feed the measured productivity back into the next work plan and budget. Rerun the spreadsheet 'what if' model with the revised resource-per-unit rates. Use the same data for staffing, equipment, workload distribution, and contract-versus-in-house decisions.

The source's activity sheet for work activity 412, Full Depth Patch, lists route segments (for example US 24 at Sterling, US 26 Rushville, Route 61 Augusta). Each segment shows survey-based needs, history, and budgeted quantities by fiscal year, with activity totals at the bottom. Productivity measurement picks up where that sheet stops. For a segment, take the quantity accomplished and divide by the man-hours or crew-days charged to it in the same period. Then compare that ratio to the standard rate used to convert budgeted quantity into resources. The source does not give the hours, so the following figures are hypothetical and for illustration only. If the budget assumed 1 unit per crew-day and a crew placed 8 units in 10 crew-days, productivity is 0.8, or 20 percent below standard. Check whether the shortfall is concentrated on one segment (a local condition or traffic factor) or spread across all of them (a standard set too high, or a crew or equipment problem). Another real precedent is the 1973 fuel shortage. Arizona, California, and other states used work programs and performance standards to trace fuel consumption by activity and to show how fuel allocations would change maintenance plans and budgets. That works because the same activity-level output and input records answer a new question.

• If actual productivity trails the standard on every location, then suspect the standard (it ignores condition, traffic, climate, or drainage) before blaming crews.
• If the shortfall appears on one segment or one crew only, then go to the field supervisor about work methods, crew and equipment arrangement, and scheduling before changing the standard.
• If output meets plan but cost per unit is above standard, then examine input mix (crew size, equipment, hours) rather than output volume.
• If you are weighing contract maintenance, then use system data on staffing levels, equipment needs, workload distribution, and work performance to compare costs, and check what a contract would do to your staffing and equipment needs.
• If a standard was built only from inventory times average effort, then supplement it with condition survey data before using it to judge productivity.

• Comparing ratios across mismatched units (man-hours in one report, crew-days in another, or different output definitions). The tell: productivity swings for no field reason when the reporting source changes.
• Judging crews against an inventory-based standard that ignores actual feature condition and local factors. The tell: the same crews look strong in good-condition areas and weak in poor-condition or high-traffic areas.
• Relying on delayed or manual records so variances are found too late to act. The tell: you learn about a problem at period-end, not while the activity is under way.
• Setting standards without field input, so supervisors discount the numbers. The tell: field explanations repeatedly contradict the reports and the standards never get revised.
• Confusing plan adherence with productivity. The tell: a crew that did less than planned is labeled inefficient, although hours per unit were on standard, or a crew that hit the quantity is labeled efficient despite overspent hours.

• ▸Activity and number: name the activity (e.g., 412 Full Depth Patch) and its code.
• ▸Location/segment and period: route segment or section and the dates covered.
• ▸Output unit and quantity accomplished: the unit used in the work plan and the actual amount completed.
• ▸Input unit and amount used: man-hours, crew-days, or cost actually charged for that output.
• ▸Actual productivity: output divided by input, computed in the same units each time.
• ▸Standard productivity: the planned rate from the work plan, with the condition, traffic, climate, and drainage adjustments you applied.
• ▸Variance and quantity gap: actual versus standard productivity, and planned versus accomplished quantity, kept separate.
• ▸Field supervisor explanation: methods, crew and equipment arrangement, or scheduling factors reported.
• ▸Action: keep the standard, revise it, or change the work method, with an owner and a review date.

1. 1Define the target first: state the condition and service level the program must deliver (e.g., signals operating normally, not on flash). Cost-effectiveness is measured against this target, not against spending alone.
2. 2Write measurable objectives, not general goals. The passages criticize 'reduce accidents and lawsuits' as a leap of faith and propose measures like 'reduce the ratio of citizen notices to county forces notices of defective traffic control devices by 20 percent.'
3. 3Add short-term operational measures (1 to 2 years) that can justify an investment even if the long-term outcome (fewer accidents or suits) does not show up.
4. 4Split activities into the three PennDOT categories: preventive (periodic checks to cut field failures), response (restore failed equipment), and design modification (fix a recurring problem). Track cost and hours separately for each.
5. 5Cost the failure side as well as the maintenance side. Count road-user costs (the Los Angeles example is in the worked example) and compare them with what proper maintenance costs.
6. 6Record real resource inputs for any new system or data base: labor, hours, travel, and quantities covered. Other agencies need this to make a sound go/no-go decision.
7. 7Capture equipment inventory, operating cost, repair cost and usage per unit so preventive maintenance and replacement decisions rest on cost data.
8. 8Link work quantities, required resources and budget line items, then use those relationships to refine standards for budgeting and work planning. Check for hidden cost exposure, such as a design change that raises contract price and cannot be absorbed within the budget year.
9. 9Review results against the original objectives on a fixed schedule and adjust the activity mix.

Los Angeles traffic signals (cited in the passages) show the cost-effectiveness test applied to the cost of failure. A 1981 report found that, despite a rigorous maintenance program, almost 24 percent of signals had malfunctions affecting traffic-flow efficiency. Rowe used the TRANSYT program to estimate that hardware malfunctions raise surface-street fuel consumption by 0.5 percent. That equals 2.8 million extra gallons of gasoline a year, about a $3.6-million annual loss to drivers. In San Francisco, nighttime crashes almost tripled when signals were put on flashing operation after midnight. The passage concludes that even a delay increase of a second or two would cost motorists far more than proper signal maintenance. The service level at stake is normal signal operation. Spending to hold it is justified because the avoided road-user cost exceeds the maintenance cost. The passages give no maintenance budget figure, so the comparison stays qualitative on the cost side.

• If a goal cannot be measured within 1 to 2 years (e.g., 'fewer lawsuits'), then add an interim operational measure before approving the investment.
• If road-user cost of poor condition (fuel, delay, crashes) exceeds the cost to maintain, then fund preventive maintenance first.
• If a problem recurs at the same site, then treat it as a design modification rather than repeating response repairs.
• If a change would raise contract price beyond what the budget year can absorb, then resolve funding before committing.
• If you cannot report the resources a system needs (hours, miles, staff), then do not claim the program is cost-effective or ask others to adopt it.

• Vague goals such as 'reduce accidents and liability'; the tell is that no baseline or target number exists to evaluate against.
• Counting only agency cost and ignoring user cost; the tell is a budget cut justified by savings while delay, fuel use or crashes go unmeasured.
• Using low-grade components or overly sophisticated configurations to save or show capability; the tell is repeated field failures and inadequate maintenance capability.
• Skipping the preventive/response/design split; the tell is that response repairs dominate and the same sites keep failing.
• Hidden implementation cost; the tell is that no one can say how many hours, miles or staff the data base took, or a mid-year change forces a contract-price increase.

• ▸Asset/activity group: name the infrastructure element covered (e.g., traffic signals).
• ▸Target condition/service level: state the measurable standard (e.g., % of signals functioning normally).
• ▸Short-term measure (1-2 yrs): give an indicator, baseline and target number.
• ▸Long-term outcome measure: give the result sought (e.g., crashes, claims) and baseline.
• ▸Cost by type: list preventive, response and design-modification cost and hours separately.
• ▸Failure/user cost: estimate delay, fuel, crash or other costs when the target is missed, with source.
• ▸Resource inputs for any new system: record staff, man-hours, miles traveled, quantities covered.
• ▸Equipment cost data: record operating cost, repair cost and usage per unit.
• ▸Decision and review date: note fund, change or stop, and when results get rechecked.

1. 1Define the condition measure before you collect anything. Pick one index (the German system's condition index, or PSI, which the passages describe as combining extent of cracking through to unevenness of the riding surface) and write down what it aggregates, so changes can be described objectively against a quality standard.
2. 2Survey by both visual and technical means. The German system builds its condition index from a visual survey plus technical measurement, not from either alone. Record each section's value at a point in time to create a database of point-in-time pavement conditions.
3. 3Categorize each section through the index, then compare the ascertained condition with the given quality requirement. The comparison, not the raw score, drives action.
4. 4Set a minimum acceptable index value. In the German approach, the decision of when to execute a maintenance measure rests on that threshold. Sections that fall below it are inadequate and move to immediate intervention (construction measures or traffic management); sections above it receive no action.
5. 5Choose a ranking approach on purpose. Blanket rating: compute one overall index per section and sort in ascending order to get a priority list. Individual rating: assess every measured defect (rutting, cracking, edge deterioration, surface stone loss, transient deflection) and place each remedy in the ordering by seriousness. Use blanket to rank the network and individual to diagnose substandard sections.
6. 6Plan deterioration as accelerating, not linear. Observations show the rate of deterioration rises with age, especially late in life, so a straight-line forecast understates how fast a section near the end of its life will fall.
7. 7Do not substitute maintenance cost for condition. Use condition data alongside the inventory-times-quantity-standard calculation so planned work reflects the actual state of the features, as Illinois, Louisiana, North Dakota and Ohio do.
8. 8Re-survey and adjust at least once or twice a year. Annual work programs are now formally adjusted mid-year, so update the condition picture on the same cycle.

The German Road and Transportation Research Association system shows the sequence. A condition survey is run by visual and technical means (checks), and a condition index is determined for each section. That index is compared with the given quality requirements. If the condition is inadequate, the system triggers immediate intervention, either by constructional measures or through traffic management. If it is adequate, priorities for the necessary maintenance measures are set, followed by planning and execution. The trigger is a minimum acceptable index value, so the decision to act is objective and repeatable instead of resting on a supervisor's judgment. In the passages' priority-rating figure, the same logic appears as bands: urgent or immediate works, a middle band funded as money is available, and sections below the minimum intervention level that get no action. For a US-style approach, you would compute PSI for every section and sort ascending to produce the priority list.

• If a section's condition index is below the minimum acceptable value, then trigger immediate intervention, either construction measures or traffic management; do not wait for the annual program.
• If you need a quick network-wide priority list, then use a blanket index sorted ascending; if a section is substandard and you must choose the remedy, then assess each individual defect.
• If a section is near the end of its useful life, then expect the deterioration rate to accelerate and shorten the re-survey interval; do not extrapolate a straight line.
• If a model assumes maintenance history has no influence on future needs, or assumes linear deterioration, then treat its output with skepticism; many maintenance engineers discount both assumptions.
• If a location shows high maintenance cost, then check condition directly before concluding it needs rehabilitation, because the cost may reflect effort that already reduced the need for maintenance.

• Using maintenance cost as a proxy for condition. The tell: a high-cost location is flagged for rehabilitation, or a low-cost one is read as sound, when the cost reflects effort already spent or a section being ignored because it is slated for resurfacing.
• Assuming straight-line deterioration. The tell: forecasts show comfortable remaining life for old pavements that then fail faster than predicted near the end of their life.
• Relying on a single aggregate index for repair decisions. The tell: a section ranks acceptably overall yet carries a serious individual defect (rutting, edge deterioration) that the blanket score masks.
• Planning from inventory times quantity standard alone. The tell: work quantities are identical for features in very different actual condition, because the equation does not consider condition.
• Treating the condition snapshot as permanent. The tell: the work program is frozen for 12 months and ignores changing conditions, even though most agencies now formally adjust once or twice a year.

• ▸Section ID and length: identify the road section being rated.
• ▸Survey date and method: record whether the data come from visual survey, technical measurement, or both.
• ▸Index used and value: name the index (condition index, PSI, PCI) and its current value.
• ▸Minimum acceptable value: state the agency threshold below which intervention is required.
• ▸Comparison result: adequate or inadequate against the quality requirement.
• ▸Individual defects present: list rutting, cracking, edge deterioration, stone loss, deflection, with severity for each.
• ▸Pavement age and life stage: note whether the section is near end of life, so you expect accelerated deterioration.
• ▸Action band: immediate intervention, funded as money is available, or no action.
• ▸Next re-survey date: set the date for the mid-year or annual review.

1. 1Define your exposure by listing the hazard types that generate suits: bumps, low shoulders, potholes, missing signs, malfunctioning signals. The passages say the causes are numerous, so exposure is broad. Start with the categories where you have data.
2. 2Choose between the two recognized techniques: risk control (minimize exposure) or risk finance (buy insurance). Many entities already carry insurance, so treat control as the half you still own. Insurance does not remove the underlying hazard.
3. 3Measure each risk on four dimensions: frequency, severity, predictability, and probability. Then pick a management technique and schedule a periodic evaluation of it. Without that evaluation, the program is a one-time effort.
4. 4Break the plaintiff's five-element chain: duty, breach, proximate cause, absence of contributory negligence, and damages. Concentrate on step two, removing negligence, because it is the ideal way to prevent court losses. Ask for each hazard whether you can show reasonable care.
5. 5Run the accident reduction program as the core of your risk management system. Make sure police know why accident data is needed and fill out reports correctly. Build a high-accident location list and look for cause patterns at each site. Develop alternative treatments, pick the most cost-effective, and rank competing sites by priority.
6. 6Erect warning signs at sites you cannot repair immediately. Review completed projects, reassess the priority list, and reassess the warning need at sites still waiting.
7. 7For signals, set repair time limits with liability in mind, using the PennDOT model. Final repair within 24 hours is required for knocked-down span wires or signal heads and for lamp burnout, conflict monitor, flasher, load switch, or signal cable failures. For other failures, emergency repair within 24 hours is acceptable, with full permit conformance within 30 days unless weather or equipment availability prevents it.
8. 8Keep a maintenance record for every complaint, in the office file and the curbside cabinet. Log who complained, when the dispatcher received it, when it went to the crew, when the crew responded, and when repair finished. Also log the trouble found (including extra problems), the repairs made, and the materials used.

PennDOT's signal maintenance rules show how to turn liability into a schedule. A span wire is knocked down in an accident. Under the rules this requires final repair, meaning the intersection is restored to proper and safe operation per the state permit within 24 hours. A different failure, one outside the lamp-burnout, conflict monitor, flasher, load switch, and signal cable list, can get emergency repair. That restores safe operation temporarily within 24 hours, and permit conformance must follow within 30 days unless weather or equipment availability prevents it. The source says picking these periods is hard because of liability implications. PennDOT's periods partly reflect that contractors doing much of the work may be far from the site. The record fields then document the timeline: complaint received, given to the crew, crew responded, repair completed. If a suit follows, the record removes doubt about what was done and when. The technician who finds an additional problem while on site logs it and fixes it, so the record shows no ignored hazard.

• If a hazard cannot be repaired immediately, then erect warning signs and keep them until the repair is complete.
• If a signal failure involves a downed span wire or head, lamp burnout, conflict monitor, flasher, load switch, or signal cable, then require final repair within 24 hours rather than emergency repair.
• If you must choose where to spend limited funds, then rank locations from the high-accident list by the most cost-effective treatment and program work in that priority order.
• If a proposed liability technique conflicts with another one, then decide at the planning level in light of local conditions. No entity can adopt every item, and some are counterproductive to each other.
• If you already carry insurance, then still run risk control, because insurance covers only the finance half of risk management.

• Treating risk management as reaction after an accident (the ostrich posture). The tell: actions follow lawsuits, not hazard lists.
• Overlooking seemingly insignificant items, which is how many cases are lost. The tell: a defect appears in a complaint that nobody logged or ranked.
• Keeping thin or office-only records. The tell: you cannot reconstruct who complained, when the crew responded, and what was repaired when litigation starts.
• Fixing only the reported fault. The tell: the technician leaves with other defects at the site unrecorded and unrepaired.
• Copying a list of techniques wholesale as a cure-all. The tell: adopted items conflict with each other and no one considered local conditions, though the source says there is no simple solution.

• ▸Location and hazard type: record the site and the defect (pothole, low shoulder, missing sign, signal malfunction).
• ▸Complainant: record who reported it.
• ▸Time complaint received by dispatcher: record the exact time.
• ▸Time given to repair crew: record when the crew was assigned.
• ▸Time crew responded: record arrival time.
• ▸Time repair completed: record completion and whether it was final or emergency.
• ▸Trouble found, including additional problems: describe all defects seen, not only the reported one.
• ▸Repairs made and materials used: list work performed and materials.
• ▸Interim warning and risk rank: note any warning signs erected, accident history, and priority rank for the site.

1. 1Define compliance on three axes before you measure it: quantity (units of each activity performed against planned), type (the right mix of activities), and timing (work done in the planned period). A program can hit its annual quantity total and still fail on timing or type.
2. 2Build the plan from the budget and the work program together. The source says work programs are still developed annually in conjunction with budget preparation, so every planned quantity should trace to a budgeted resource level.
3. 3Keep the activity list small enough to track. Some pre-1980 systems defined more than 100 activities, and some described almost 400 kinds of work. Oklahoma DOT uses less than half of its original 125, and many new systems define fewer than 50. Consolidate until each activity has a plan quantity, a timing window and an actual-quantity record.
4. 4Attach a quantified level of service to every field activity, as Caltrans is doing. This links staffing and budget to what the work buys, so a planned quantity is a decision with a known consequence. Management can then see what it gets and what it forgoes when staffing moves up or down in a subprogram.
5. 5Allocate resources geographically with the same level-of-service logic, and issue an annual work plan to each field manager. A field manager can only be held to a plan that came from an equitable, explainable allocation.
6. 6Compare actual to plan on a fixed cycle and run formal field compliance reviews. Caltrans' quantified levels of service are meant to be the basis for those reviews. Report variance by activity, by location and by period.
7. 7Schedule one or two formal mid-year adjustments. Most programs now get them, and county and local agencies also face mid-year budget changes. Record each change as a revision to the plan so compliance is measured against the current approved plan, not a stale one.
8. 8When utilization is low, test the plan before blaming the field. The source notes that low utilization can result from a work program that requires far fewer units (equipment or crews) than are on hand.

The source gives two situations where a good manager departs from the annual plan. In the first, a growing season calls for one more mowing than planned. In the second, an early fall curtails the planned amount of preventive maintenance. Say you plan N mowing cycles and a long season requires N+1. Measured against the original plan, the district shows a 1-cycle overrun on mowing. Measured against a formally adjusted plan, it is compliant. The early fall cuts preventive maintenance below plan, so you document the weather cause and move the shortfall into the mid-year revision. The source treats formal once-or-twice-yearly adjustments as the norm, and says they make the program more effective and more responsive to changing needs. The rule is that departures are acceptable when they are formal and documented. A silent drift from the plan is not.

• If a variance traces to weather or a natural disaster (an extra mowing, a curtailed fall preventive program), then revise the plan formally at the next adjustment and score compliance against the revised plan.
• If equipment or crew utilization is low, then check whether the work program requires far fewer units than are on hand before treating it as a field performance problem.
• If your activity list runs to hundreds of items and variance cannot be explained by activity, then consolidate toward the Oklahoma pattern (under half of the original list, often under 50) so each activity is measurable.
• If management proposes raising or lowering staffing in a subprogram, then state the quantified levels of service that change with it before approving the budget.
• If you adopt a written standard or guideline, then keep it current and attainable. The liability passage warns that failing to adhere to an adopted standard can constitute negligence.

• Treating the annual plan as fixed for 12 months. The tell is that compliance reports show large variances for weather or midyear budget changes, and nobody has revised the plan.
• Over-defining activities. The tell is a list of 100 to 400 work types where field crews code work loosely and plan-versus-actual comparisons are not credible.
• Measuring quantity only. The tell is that annual totals match the plan while the work happened in the wrong season, for example preventive maintenance squeezed into a shortened fall.
• Setting plan quantities with no level-of-service basis. The tell is that managers cannot say what is lost when budget or staffing drops, and compliance reviews turn into arguments about whether the target was fair.
• Adopting standards that cannot be met. The tell is a gap between written standards and actual performance, which creates negligence exposure when the standard is not followed.

• ▸Activity and subprogram: name the activity and the subprogram it belongs to.
• ▸Level of service: state the quantified level of service the plan quantity is meant to deliver.
• ▸Planned quantity and timing window: enter the units and the period from the approved annual work plan.
• ▸Actual quantity and timing: enter the units completed and the period in which they were done.
• ▸Variance (quantity and timing): record the difference from plan for each, kept separate.
• ▸Cause of variance: mark weather or disaster, budget change, resource shortage, planning error or field performance.
• ▸Formal plan revision needed (Y/N, date): note whether the variance justifies a mid-year adjustment and when it is logged.
• ▸Utilization check: compare units on hand to units the program requires, to flag a work program that is too small for the resources.
• ▸Field manager and review date: record who owns the plan and when the compliance review took place.

1. 1Combine the financial budget and the performance budget into one format. Each activity line carries four things: a description of the activity, the planned service level and cost, the resource need, and the proposed budget by object of expenditure. The passages' striping example: 'Pavement line striping performed annually or as needed to delineate traffic lanes and thereby promote traffic safety.' It is planned at '3,950 miles of lane line and pavement edge striping - $643,000.'
2. 2Break each activity's planned quantity into resources and line items. For every object-of-expenditure code (e.g., '101-320-002'), show the total current and proposed quantities of people, equipment, materials and services. Show last year's actual expenditure, the current budget, and the proposed budget. The Mtce. Worker II line reads 1.2 / 1.2 with $19,872, $20,050 and $20,610.
3. 3Move to single-source reporting. The individual or crew reports all maintenance management and cost-related data daily in one document. Labor, equipment, and materials used go on that one form. The passages' PennDOT example plans a combined form showing man-hours, materials, and equipment, so the foreman codes a day's information once.
4. 4Translate the reported resources into dollars automatically. Use resource cost tables that are updated daily, or access the regular accounting ledgers. Do not re-key the quantities by hand into the accounting system.
5. 5Modify the chart of accounts so it accepts maintenance activity codes. Relate each activity to its sources of revenue, objects of expenditure, or account elements. Without this step, activity costs and ledger entries cannot be reconciled.
6. 6Modify materials accounting so materials are costed out as they are used, not when the invoice is paid. Expand the traditional inventory account and use inventory clearing accounts so activity costs reflect actual consumption.
7. 7Store actual quantities and actual dollars in a common data base. Sort the dollars to meet user needs: the manager sees activity performance and cost, and fiscal control sees expenditures. Test the integration by checking that the accounts balance to the dollar.
8. 8If your management, inventory, and equipment systems are separate computer systems (as in PennDOT, where they were developed at different times), plan an integrated data base with a common input form. Use formulas based on established work standards to recalculate resource requirements whenever work units change.

Take the striping line in the passages' budget format. The performance side states the activity: pavement line striping, performed annually or as needed to delineate traffic lanes. The planned service level is 3,950 miles of lane line and pavement edge striping at $643,000. The resource side lists object-of-expenditure lines. One is code 101-320-002, Mtce. Worker II, at 1.2 positions in both current and proposed quantities. It shows $19,872 last year's actual, $20,050 current budget, and $20,610 proposed. During the year the crew reports labor, equipment, and materials on one daily document. Automated resource cost tables convert those quantities to dollars. The modified chart of accounts books the dollars to the striping activity code and to the right object of expenditure. Materials are costed as used through inventory clearing accounts. At year end the actual dollars sorted by activity and by object of expenditure must balance to the dollar, and they feed next year's budget.

• If maintenance and accounting data are reported separately, then adopt single-source daily reporting before building any budget format, because the accounts will not balance without it.
• If materials are charged only when invoices are paid, then modify materials accounting so costs are recorded as the materials are used.
• If the chart of accounts has no maintenance activity codes, then refine it to relate each activity to revenue sources and objects of expenditure before linking budgets.
• If the management, equipment, and inventory systems are independent computer systems, then build an integrated data base with a common input form rather than reconciling reports by hand.
• If a work quantity changes in the plan, then recalculate resource requirements from work standards and revise the dollar request in the same pass.

• Keeping the performance budget and the financial budget as separate documents. The tell: the work plan total and the fiscal request disagree and need manual reconciliation.
• Costing materials at invoice payment. The tell: activity costs spike or lag against actual work done, and the inventory account does not clear.
• Adding activity codes to the field forms but not to the chart of accounts. The tell: crews report by activity while finance reports only by object, and the two never tie out.
• Leaving separate systems (management, equipment, inventory) with separate input forms. The tell: foremen code the same day's hours and materials on more than one form.
• Treating the balance as optional. The tell: accounts do not balance to the dollar, so the budget cannot be trusted as realistic.

• ▸Activity description: one-line statement of the work and its purpose
• ▸Planned service level / annual work quantity: units and amount (e.g., miles)
• ▸Total funding requirement: dollars tied to that quantity
• ▸Object of expenditure line item code: the account code each resource charges to
• ▸Resource quantity (current / proposed): people, equipment, materials, services
• ▸Last year's actual / current budget / proposed budget: three dollar values per line
• ▸Source of revenue or account element: the funding source the activity charges to
• ▸Daily reporting document used: the single form on which labor, equipment, materials are reported
• ▸Balance check: whether activity-sorted actual dollars equal ledger dollars (yes/no, variance)

1. 1Define the five uses before buying anything: maintenance planning, data entry and editing, inventory management, 'what if' analysis, and timely reports. Rank them by the decision each one supports at the district or local office.
2. 2Start with data entry at the district. The passages describe a move from centralized data entry and batch processing to remote terminal entry of planned and actual values, handled interactively. Put the entry and editing point where the work is reported, and enter planned and actual values together.
3. 3Build the feature inventory next, because the other uses depend on it. Clinton County, Ohio built a computerized highway inventory system, and some programs update feature inventories automatically. Keep the inventory procedure simple, but do not drop it.
4. 4Program the 'what if' layer on the work plan. The programs recalculate maintenance work plans, budgets, monthly work load distributions, and resource needs for actual or projected circumstances. Test it with one scenario that used to take hours by hand, and confirm it answers in seconds.
5. 5Store work quantity and expenditure data by activity and general location. This is the history you use to project future maintenance needs and costs.
6. 6Set up reports on actual work accomplished, productivity, unit cost, and total cost by activity, available as soon as files are updated. The goal is day-by-day monitoring of an activity or project without a separate manual record.
7. 7If the micro links to a mainframe, decide which tasks stay local and which go up. Use the micro for interactive 'what if' work and entry, and the mainframe for the shared data base.
8. 8Assess the contribution concretely, as Brewer challenges managers to do, office by office and for the overall system. Write down what changed in each office, and do not assume the machine improved management, communication, or deadlines.

The passages name Clallam County, Washington, and Little Rock, Arkansas, as local agencies that moved furthest. Their microcomputers, or micros linked to mainframes, answer 'what if' questions in seconds rather than hours. Suppose a county office faces a change in the projected work load. The program recalculates the work plan, the budget, the monthly work load distribution, and the resource needs together. Some of these programs also update the feature inventory and create and monitor work orders. When the files are updated, the manager reads actual work accomplished, productivity, unit cost, and total cost by activity. That makes day-by-day monitoring possible with no separate manual set of records. The same system-generated data on staffing, equipment, and work load can test a contract-maintenance proposal. It can show how a contract would change staffing and peak-season work loads. Brewer's caution still applies: the machine only speeds up processing, and it does not fix weak management.

• If the office still re-keys data at headquarters or runs batch cycles, then move entry and editing to the district first, because turnaround time is what makes daily monitoring possible.
• If a 'what if' answer still takes hours, then the program does not recalculate the plan, budget, work load distribution, and resources together; fix that before adding reports.
• If a manager lacks communication or planning discipline, then do not expect a microcomputer to supply it; fix the practice, because Brewer states that a micro will not make a bad manager good or make late people submit early.
• If a use cannot be tied to a decision made at the local level, then do not adopt it; Brewer's test is a concrete assessment of what each office gains.
• If a proposed use is experimental rather than a settled trend, then pilot it in one office before rollout.

• Expecting the machine to fix management or communication. The tell: reports are no earlier, and staff use the extra processing power to do more before the deadline rather than finish sooner.
• Adopting hardware without a shared data base or integration. The tell: the micro duplicates records that headquarters also keeps, and totals do not reconcile.
• Simplifying the inventory or defined activities so far that the system cannot support analysis. The tell: you cannot compute unit cost by activity or project needs.
• Keeping a manual parallel record. The tell: staff still maintain separate paper logs, so the speed advantage in timely reporting never appears.
• Treating experimental configurations as proven trends. The tell: the passages call many hardware and software setups experiments, yet the setup is rolled out widely with no pilot evidence.

• ▸Office or district: name the local unit being assessed.
• ▸Maintenance planning use: state whether work plans and budgets are built on the micro, and how.
• ▸Data entry and editing: say where planned and actual values are entered and whether it is interactive or batch.
• ▸Inventory management: record whether feature inventories are on the micro and how they are updated.
• ▸'What if' capability: give the time to recalculate a scenario (seconds or hours) and which items update.
• ▸Reports produced: list performance and cost reports (work accomplished, productivity, unit cost, total cost) and how soon after file updates.
• ▸Link to mainframe: note whether the micro is stand-alone or linked, and which tasks sit where.
• ▸Concrete contribution: describe in one line what changed in this office, supported by evidence.
• ▸Next action and owner: name the single next step and who is responsible.

1. 1Build the data base first. Collect the long-term physical and geometric characteristics of each road section plus its current condition. Snaith and Burrow state that considerable data on both are required before remedial works can be selected. Where the old 'inventory times quantity standard' was used, add field condition surveys so work quantities reflect actual condition, as Illinois, Louisiana, North Dakota and Ohio do.
2. 2Sort every section into three bands before ranking (Snaith and Burrow, Figure 1). (A) Reconstruction candidates, which fall outside normal maintenance priority. (B) Urgent or immediate faults, such as blocked drains and potholes that lead to rapid, progressive failure. (C) Sections below the minimum intervention level, which get no action. Rank only the middle population.
3. 3Choose a rating basis. A blanket rating, such as the overall PSI per section sorted ascending, is simple. An individual-item rating scores each measured defect (rutting, cracking, edge deterioration, surface stone loss, deflection) and places the remedy by seriousness. The source prefers the individual approach, because a section with many minor faults should not outrank a section with one extremely serious defect.
4. 4Separate user comfort from structural integrity. A very bumpy road can be structurally sound, and a smooth one can be close to failure. Rate both, and state in writing which one governs when they conflict.
5. 5Cost each item. Snaith and Burrow note costing is needed so the engineer can choose the cut-off point in the priority list. Attach an estimated cost to every remedial item, then total down the ranked list.
6. 6Draw the cut-off line where cumulative cost meets the available money. Items above the line are funded, and items below it are carried forward or documented. Show the line on the list so decision-makers see what is not being done.
7. 7Record the engineering judgment as explicit rules. The maintenance management system requires judgment to be set out as a series of rational decisions it can implement. Write down each override, such as 'fix blocked drains on observation', so it can be audited.
8. 8Reconcile cost data with accounting. Maintenance managers and accountants often disagree on spending by a 4 to 7 percent spread. Use one cost source for the priority costing, so rankings are not argued over dollar differences.

Take the Snaith and Burrow framework applied to a network with too little money. Section 1 is badly broken and needs complete reconstruction, so it is pulled from the maintenance list and flagged. Section 2 has a blocked drain and a pothole, which lead to rapid and progressive failure, so they go straight to immediate works on observation. Section 3 has many minor faults, and Section 4 has one extremely serious defect. A blanket PSI sort might favor Section 3, but the individual-item rating puts Section 4 first because of the defect's seriousness. Section 5 has no observed fault, so it falls to the lowest level and gets no action. The remaining items are costed and totaled down the ranked list, and the cut-off is drawn where the available budget runs out. Items below the line are deferred and visible, not silently dropped.

• If a section needs complete reconstruction because accepted maintenance treatments are insufficient, then remove it from the maintenance ranking and flag it separately.
• If a fault will lead to rapid and progressive failure (blocked drains, potholes), then treat it as immediate works on observation rather than ranking it with routine items.
• If one section has many minor faults and another has one extremely serious defect, then rank the serious-defect section higher; use individual-item rating, not a summed blanket score.
• If comfort and structure disagree (bumpy but sound versus smooth but weak), then apply your documented policy on which governs, and do not let the ride-quality number decide by default.
• If the ranked, costed list exceeds available funds, then set the cut-off where cumulative cost meets the budget and report what falls below it.

• Ranking by worst measured condition alone. Snaith and Burrow warn there is more to priority assessment than placing the worst condition at the top. The tell: the list ignores traffic, cost and failure progression.
• Using one blanket index (PSI) for everything. The tell: sections with many minor defects outrank sections with a single severe one.
• Letting unobserved sections drop out silently. A section with no observed fault is automatically placed at the lowest priority. The tell: a poor section ranks low only because it was never surveyed.
• Leaving judgment implicit. The tell: the system cannot reproduce why an item was moved, because engineering judgment was not set out as rational decision rules.
• Ranking without costing, or with disputed cost data. The tell: no cut-off point can be defended, or finance and maintenance disagree on spend by 4 to 7 percent.

• ▸Section ID and location: identify the road section being rated.
• ▸Band (A reconstruct / B immediate / C below minimum intervention / ranked): assign before any scoring.
• ▸Measured defects: list each defect (rutting, cracking, edge deterioration, stone loss, deflection) with its measured value.
• ▸Most serious single defect and its severity: this drives the individual-item rating.
• ▸Overall PSI (if used): record as a cross-check, not the sole ranking basis.
• ▸Comfort versus structural flag: mark whether the issue is ride quality, structural risk, or both.
• ▸Proposed remedial work: state the treatment selected for the item.
• ▸Estimated cost and cumulative cost: enter the cost, then the running total down the ranked list.
• ▸Cut-off decision: funded, deferred, or no action, with the budget line used and the judgment rule applied.

1. 1Pull system-generated data before you decide: staffing levels, equipment needs, work load distributions, and work performance records for the activity under review. The source treats this data as the basis of a contract decision, not an afterthought.
2. 2Run the cost test first. Ontario's strategic policy is to contract only where financial analysis and assessment indicate total cost will be reduced, using maintenance management system costs as the input. Compare in-house cost per unit against the contract price on the same basis.
3. 3If cost does not decide it, test for expertise. The source accepts contracting on the availability of expertise or special licenses in the private sector that the agency could not warrant keeping in-house. Record which skill, license, or equipment is missing.
4. 4Model the effect on the whole maintenance operation by answering four questions: What happens to staffing levels and equipment needs? What happens to current work load distributions? Does the contract reduce peak-season work loads or create slack-season problems? Can the contractor use standard work methods, and can the standard help define the work?
5. 5Scope at the activity level first. In Ontario, contractors supply sand spreaders, plow trucks, surface treatment, gravel crushing, hot-mix patching, and stockpiling. Total maintenance of a highway section had not been contracted out. Start with discrete activities unless you have a reason to go further.
6. 6Use the performance standard to write the work definition. If the standard defines the work, it becomes the contract specification, and you can price the activity against it.
7. 7Put cost monitoring in place before award. Ontario's shift to contracting created a need for a comprehensive maintenance management system to monitor private operations and ensure least-cost alternatives are chosen.
8. 8Re-run the analysis periodically. Contracted work that was once in-house (mowing, sweeping, catchbasin cleaning, brush clearing) should be re-checked against current costs, since the system data keeps accumulating.

Ontario's Ministry of Transportation and Communications (MTC) shows the pattern. Contracting for parts of the highway system began in the 1950s and accelerated in recent years, especially in surface treatment, gravel crushing, hot-mix patching, and winter maintenance. Almost all sand spreaders are now mounted on contractors' trucks, and about 23 percent of snowplow trucks are supplied and operated by contractors. Large urban contracts cover stockpiling and spreading winter sand and salt. Most surface treatment, liquid calcium chloride application, crushed gravel supply, dome construction, and picnic site maintenance is contracted, along with some mowing, sweeping, catchbasin cleaning, and brush clearing. The Ministry's rule is to contract where financial analysis, using maintenance management system costs, indicates total cost will fall. Expertise or special licenses unavailable in the department are the second accepted basis. Total maintenance of whole highway sections has not been contracted out. The shift forced reorganization of activities and a system to monitor contractor costs and confirm least-cost selection.

• If analysis using maintenance management system costs shows total cost falls under contract, then contract the activity. If it does not, keep it in-house unless a skills or licensing gap applies.
• If the activity needs expertise or a special license the department could not justify holding, then contract it even without a clear cost advantage.
• If the contract would deepen slack-season problems or leave staff and equipment idle, then reconsider or restructure it. If it reduces peak-season work loads, that counts in favor.
• If the standard work methods cannot define the work in a form a contractor can follow, then fix the standard before contracting.
• If you contract more than a few activities, then build cost monitoring first, because the Ontario experience required a system to ensure least-cost alternatives.

• Deciding on price alone and ignoring staffing and equipment effects. The tell: idle crews or surplus equipment after award, and no analysis of work load distribution on file.
• Contracting without a defined standard. The tell: disputes over what counts as completed work, and no link between the contract and the performance standard.
• Creating slack-season problems by pulling out only the peak-season work. The tell: in-house crews underused part of the year while contract costs are paid at the peak.
• Skipping cost monitoring after award. The tell: no one can say whether the contractor is actually the least-cost alternative.
• Treating the decision as permanent. The tell: activities contracted years ago have never been re-analyzed against current in-house cost data.

• ▸Activity and unit of measure: name the work (e.g., surface treatment, mowing) and how it is measured.
• ▸In-house cost per unit: from maintenance management system cost records.
• ▸Contract price per unit: bid or quoted price on the same basis.
• ▸Total-cost difference: include effects on staffing and equipment, not just unit price.
• ▸Skill, license, or equipment gap: list any that the department could not warrant holding.
• ▸Staffing and equipment impact: state what changes in crews and equipment needs.
• ▸Work load distribution impact: does it reduce peak-season load or create slack-season problems?
• ▸Standard work method defined?: yes/no, and whether the standard can serve as the work definition.
• ▸Decision and review date: contract, in-house, or split, with the date to re-run the analysis.

1. 1Define the purpose before you collect: state which decisions the data will feed (work quantities by activity, priority lists, rehabilitation strategy, liability defense). The passages show condition surveys supplementing the 'inventory times quantity standard' calculation, which ignores actual feature condition.
2. 2Build the inventory first. Identify and locate every maintainable item, then gather data on each. Clinton County, Ohio began with 270 miles of county roads, 290 bridges, 1,100 culverts, and 2,300 signs spread across county, township, and state road systems.
3. 3Enter inventory and condition data into one computerized data base so records are consistent and can be re-sorted. Clinton County's phase (c) did this.
4. 4Add a condition survey, visual or mechanical, that records point-in-time condition by route section or organizational unit. Summarize it into totals by activity so planned work quantities reflect actual condition, as Illinois, Louisiana, North Dakota, and Ohio do.
5. 5Categorize conditions through a consistent index, then rank deficiencies into a priority list. Cost each item so you can pick a cut-off point within budget. Do not just put the worst measured condition at the top: the Snaith and Burrow paper warns that priority assessment involves more than that.
6. 6Match inventory detail to what the decision needs. A traditional inventory counts or measures every feature, and pavement management systems demand even more surface detail. The passages cite research suggesting full counts may be unnecessary, so test whether a simplified procedure is enough.
7. 7Close the loop: add new data as items are built or changed, devise reporting forms, and perform periodic inspections (Clinton phases e and f). Keep records both in office files and, for signals, in the curbside cabinet, since adequate records reduce liability.
8. 8Revisit the resulting work program at least once or twice a year with formal adjustments. Annual programs no longer hold for 12 months, and midyear budget changes require refinements.

Clinton County, Ohio, a rural county with local roads under 500 ADT and many single-vehicle accidents, needed to know whether roadside features and traffic control devices met state and federal guidelines. It followed six phases: (a) identify and locate items, (b) gather data on them, (c) enter data into a computerized data base, (d) establish priorities, (e) add new data, and (f) devise reporting forms and perform periodic inspections. The scope was 270 miles of county roads, 290 bridges, 1,100 culverts, and 2,300 signs on county, township, and state systems. From the data base, the county produced a priority list for systematic correction of deficient conditions within current budget restraints. This mattered because an Ohio Supreme Court ruling eliminated sovereign immunity, and budget limits are often not accepted as a reason to skip corrective measures. Showing reasonable intent to correct problems systematically can reduce liability risk. The long-term goal was fewer accidents from faulty or substandard features and fewer lawsuits.

• If your work quantities come only from inventory times a standard, then add a field condition survey so total quantities by activity and route section reflect actual condition.
• If a pavement model assumes straight-line deterioration, then distrust it: observations show deterioration accelerates with pavement age.
• If maintenance cost data for a segment looks negligible, then check whether the segment is solid, or is being ignored because it is slated for rehabilitation, before reading it as low need.
• If budget cannot fund every deficiency, then rank items by priority, cost them, and set a documented cut-off, since budget limits alone are not a defensible reason for inaction.
• If your features include roadside, sign, and drainage work, then do not rely on pavement models alone. They cover paved surfaces and miss work that can exceed half of an agency's workload and budget.

• Treating inventory as condition: planned work is computed from feature counts and averages, so the plan ignores actual condition. The tell: identical quantities for sections in very different shape.
• Adopting a pavement model without checking validity: straight-line deterioration assumptions underpredict late-life decline. The tell: predictions that diverge from observed pavement behavior.
• Collecting pavement data only: the model covers paved surfaces while other features take over half the effort. The tell: no condition data for signs, drainage, or roadside.
• Ranking purely by worst measured condition: priority lists ignore other parameters the Snaith and Burrow paper flags. The tell: the top of the list is not what engineers consider most pressing.
• Over-detailed or stale data: full counts of every feature that decisions don't need, or a data base never updated. The tell: records lag the field, and no one adds new data or runs periodic inspections.

• ▸Decision supported: name the decision (work quantities, priority list, rehab strategy) this data feeds.
• ▸Feature classes inventoried: list pavement, bridges, culverts, signs, roadside, drainage, and any others, with counts.
• ▸Location identifier: state how each item is located (route section, organizational unit).
• ▸Survey method per class: visual or mechanical, and the detail level justified by the decision.
• ▸Condition index/categories: define the rating scale used to categorize each item.
• ▸Data base location and owner: where data is entered and who maintains it.
• ▸Priority and cost basis: how items are ranked and costed, and the budget cut-off point.
• ▸Update and inspection schedule: when new data is added and periodic inspections occur.
• ▸Records retention: what is kept in office files and in the field cabinet, for liability.