Capacity and Utilization

100% utilization is not a trophy

Heat treat ran at 96% utilization while OTD collapsed. Utilization rose because queues grew; the metric looked green while customers reddened. Capacity is maximum sustainable output; utilization is actual over capacity.

FlowForge Components is a precision parts supplier to automotive and aerospace OEMs and the anchor organization for OPS 201. Annual revenue is approximately $215M. OEE (overall equipment effectiveness, the product of availability, performance rate, and quality rate for equipment) runs near 78% across 42 CNC machining centers. External defect rate is 1.2% on shipped lots. VP Operations Nina Kowalski and Plant Manager Greg Santos lead process capacity, quality systems, and lean operations across 3 plants: Toledo (main campus, 520 staff), Monterrey machining (210 staff), Cleveland finishing and CMM (110 staff).

Every lesson ties frameworks to FlowForge decisions: capacity investments, quality escapes, lean waste removal, and demand forecasts that feed master schedules. You should finish each lesson able to explain the topic to a smart colleague who has not taken OPS 201, using reconciled numbers where the topic requires arithmetic.

Process capacity and the bottleneck

Process capacity equals the minimum capacity across serial steps (unless parallel). FlowForge heat treat: 1,800 parts/day; CMM effective 2,200; machining 2,600. Bottleneck: heat treat at 1,800.

Utilization and its traps

Utilization = throughput ÷ capacity. Near 100% at the constraint with variability creates exploding queues (Kingman formula intuition). Protective capacity is strategic.

Effective versus theoretical capacity

Subtract changeovers, maintenance, and yield loss. Greg plans on effective capacity; sales often quotes theoretical.

Economies of scale and diseconomies

Scale helps until coordination cost and congestion dominate. FlowForge diseconomy appears past 92% constraint utilization without WIP control.

Shift pattern 3×8; 1,800 effective parts/day.

Part A: Resources

Part B: Implied max process

Process cap = min = 1,800/day

Part C: Protective capacity

Target constraint util 85%; releases 162 parts/day equivalent slack. Check: queue hours 38→19 ✓

Part D: Managerial read

Do not celebrate 96% util; cap WIP and add 8% heat treat capacity.

Fictional plant ran machining overtime while heat treat idled parts; util looked busy, output flat.

Machining cap 2,600; demand 2,450; heat treat 1,800; demand same. Tasks: (1) Bottleneck? (2) Max sustainable system output? (3) Machining util?

Solution

Bottleneck heat treat 1,800. System max 1,800/day. Machining util 1800/2600=69.2%. Check: matches process cap ✓

Key takeaways

• Process capacity is set by the bottleneck step.
• High constraint utilization without WIP control destroys OTD.
• FlowForge heat treat at 96% util was a warning, not success.
• Plan with effective capacity, not theoretical nameplates.
• Protective slack at the constraint buys dependability.

1. Find bottleneck in a process you know.
2. Why did 96% utilization hurt FlowForge OTD?
3. Continue to Lesson 4: Little's Law in Practice.

Applying Capacity and Utilization at FlowForge scale

When FlowForge Components evaluates capacity and utilization, VP Operations Nina Kowalski and Plant Manager Greg Santos start from operational facts: $215M revenue, 78% OEE (overall equipment effectiveness), 1.2% external defect rate, and 94% on-time delivery to OEM customers. The process analysis, flow metrics, and capacity review cadence is weekly on the Toledo shop floor and monthly with the CEO and CFO. A lesson concept that sounds abstract becomes concrete when tied to CNC cycle times, heat-treat queue lengths, and PPAP (production part approval process, the automotive quality gate before volume shipment) holds.

Consider how a one-point OEE improvement affects FlowForge. At 42 machining centers running three shifts, a single point of OEE often frees roughly $1.8M to $2.4M of effective capacity annually without new capital, depending on bottleneck mix and scrap rework rates. That is why capacity and utilization is not academic for Nina Kowalski; it is how the company funds automation without missing aerospace delivery windows.

The process analysis, flow metrics, and capacity workflow at FlowForge deliberately separates descriptive dashboards from causal improvement tests. A spike in WIP (work in process, partially completed units between operations) triggers a value-stream walk before overtime is approved. A quality escape triggers containment, root-cause analysis, and SPC (statistical process control, using control charts to distinguish common-cause from special-cause variation) review on the affected line. Forecast errors trigger aggregate-planning revisions before raw bar stock is purchased. Label outputs before they reach the executive committee: observation, tested mechanism, or scaled policy.

Document definitions alongside every operations metric tile. FlowForge's OEE formula specifies availability losses (planned maintenance versus breakdown), performance losses (speed versus standard cycle), and quality losses (scrap and rework at the constraint). On-time delivery excludes customer-approved pull-ins but includes contractual grace days. Defect rate is measured at OEM incoming inspection per million opportunities. When definitions live in a shared dictionary, the company builds institutional memory instead of re-debating the same spreadsheet every quarter.

Extended FlowForge scenario: cross-functional read

Imagine FlowForge's Q3 review for capacity and utilization. Finance asks whether a capacity investment clears hurdle rate given 8.2 inventory turns and rising interest expense. Commercial asks whether on-time delivery can hold at 94% if automotive mix shifts toward shorter lead-time programs. Quality asks whether the 1.2% external defect rate threatens PPAP status on a new aerospace cell. A weak process analysis, flow metrics, and capacity answer addresses only one function. A strong answer shows how evidence flows: process maps localize WIP buildup at heat treat, capacity models quantify constraint hours, control charts separate noise from special cause, and forecast error bands drive staffing and inventory buffers.

Work the arithmetic on a conservative example. Suppose FlowForge's heat-treat line processes 1,800 parts per day at the constraint while downstream CMM inspection can clear 2,200 units per day. Increasing heat-treat throughput 8% without adding inspection capacity may only relocate the bottleneck and inflate WIP. Multiply queue delay by average margin per part to communicate dollar risk to executives who do not live in Gantt charts. Pair point estimates with guardrails: scrap rate, overtime hours, and customer premium freight.

Stakeholder conflict is normal. Greg Santos may push overtime to clear a automotive backlog while Nina Kowalski holds spending until lean kaizen (continuous small improvements, Japanese for "change for the better") tests finish. The CFO may push inventory cuts that lengthen setup-heavy campaigns. Capacity and Utilization gives you language to negotiate those tensions with capacity, quality, and forecast evidence rather than charisma.

Translate lessons to your own context by replacing FlowForge names while keeping structure. Pick one operations decision you face this quarter. Write the process boundary, constraint assumption, primary metric, guardrails, and kill criteria before changing the schedule. If you cannot write those elements, you are not ready to approve overtime or capital regardless of how urgent the email thread feels.

Technical mechanics and checks (worked patterns)

For capacity and utilization, FlowForge analysts show work the way finance shows reconciliations. A process capacity table lists resource, time per unit, units per hour, daily capacity at stated shift pattern, and a check that the bottleneck matches the lowest capacity step. A Little's Law table prints average WIP, throughput, and implied flow time with a check that $I = R \times T$ reconciles within rounding. A control-chart appendix lists subgroup size, center line, control limits, and rule violations before a line stop is authorized. A forecast table shows actual, forecast, absolute error, and cumulative bias by family.

Use plain-language statements before formulas. Example for capacity: process capacity equals the minimum capacity across serial steps unless parallel paths merge. FlowForge forbids ambiguous one-word metrics like efficiency without stating whether it means OEE, labor efficiency, or first-pass yield. Each definition implies different data collection and different managerial meaning.

For spreadsheet or ERP replication, write the grain first. Order-line tables suit on-time delivery. Operation-sequence tables suit routing-based capacity. Shift-level tables suit OEE losses. SKU-family tables suit forecast accuracy. FlowForge Components ties every lesson metric to a named owner on the operations review slide.

Common executive questions (and disciplined answers)

Executives ask short questions that require long disciplined answers. "Are we capacity constrained?" maps to bottleneck utilization, WIP shape, and overtime trend, not gut feel from the parking lot. "Is quality getting better?" maps to defect Pareto, SPC signals, and cost of poor quality, not one good week after a customer audit. "Can we trust the forecast?" maps to bias, MAPE (mean absolute percentage error), and forecast value added versus a naive baseline. "Why not just add a shift?" maps to demand permanence, training cost, and whether the constraint moves.

FlowForge's credible answer format for capacity and utilization is three bullets: recommendation, evidence strength (descriptive, tested, scaled), and next study if limitations matter. A fourth bullet lists what would falsify the recommendation within sixty days. That discipline prevents the operations team from becoming either a bottleneck or a rubber stamp.

Linking Capacity and Utilization to prior and next lessons in OPS 201

Operations fluency is cumulative. Capacity and Utilization in Unit 2 connects backward to definitions and forward to integrative decisions. When you read FlowForge examples, mark which numbers are structural (routing standards, shift calendars, contractual service levels) versus policy (safety stock targets, overtime triggers, inspection sampling rates). Mixing the two produces recommendations that work once and fail next quarter.

Nina Kowalski's team keeps a single-page operating system for each plant: strategic priorities from Unit 1, process facts from Unit 2, service and queue policies where customers wait, quality and lean cadence from Unit 4, planning horizons from Unit 5, and capital or outsourcing choices from Unit 6. Capacity and Utilization should slot into that page with an owner and review frequency. If it does not slot anywhere, it is trivia.

Practice teaching capacity and utilization aloud using only FlowForge nouns and one table. If your explanation requires generic "a factory," you have not yet transferred the lesson. Retry with 1,800 parts per day, 78% OEE, and a named OEM program deadline.

Applying Capacity and Utilization at FlowForge scale

When FlowForge Components evaluates capacity and utilization, VP Operations Nina Kowalski and Plant Manager Greg Santos start from operational facts: $215M revenue, 78% OEE (overall equipment effectiveness), 1.2% external defect rate, and 94% on-time delivery to OEM customers. The process analysis, flow metrics, and capacity review cadence is weekly on the Toledo shop floor and monthly with the CEO and CFO. A lesson concept that sounds abstract becomes concrete when tied to CNC cycle times, heat-treat queue lengths, and PPAP (production part approval process, the automotive quality gate before volume shipment) holds.

Consider how a one-point OEE improvement affects FlowForge. At 42 machining centers running three shifts, a single point of OEE often frees roughly $1.8M to $2.4M of effective capacity annually without new capital, depending on bottleneck mix and scrap rework rates. That is why capacity and utilization is not academic for Nina Kowalski; it is how the company funds automation without missing aerospace delivery windows.

The process analysis, flow metrics, and capacity workflow at FlowForge deliberately separates descriptive dashboards from causal improvement tests. A spike in WIP (work in process, partially completed units between operations) triggers a value-stream walk before overtime is approved. A quality escape triggers containment, root-cause analysis, and SPC (statistical process control, using control charts to distinguish common-cause from special-cause variation) review on the affected line. Forecast errors trigger aggregate-planning revisions before raw bar stock is purchased. Label outputs before they reach the executive committee: observation, tested mechanism, or scaled policy.

Document definitions alongside every operations metric tile. FlowForge's OEE formula specifies availability losses (planned maintenance versus breakdown), performance losses (speed versus standard cycle), and quality losses (scrap and rework at the constraint). On-time delivery excludes customer-approved pull-ins but includes contractual grace days. Defect rate is measured at OEM incoming inspection per million opportunities. When definitions live in a shared dictionary, the company builds institutional memory instead of re-debating the same spreadsheet every quarter.

Extended FlowForge scenario: cross-functional read

Imagine FlowForge's Q3 review for capacity and utilization. Finance asks whether a capacity investment clears hurdle rate given 8.2 inventory turns and rising interest expense. Commercial asks whether on-time delivery can hold at 94% if automotive mix shifts toward shorter lead-time programs. Quality asks whether the 1.2% external defect rate threatens PPAP status on a new aerospace cell. A weak process analysis, flow metrics, and capacity answer addresses only one function. A strong answer shows how evidence flows: process maps localize WIP buildup at heat treat, capacity models quantify constraint hours, control charts separate noise from special cause, and forecast error bands drive staffing and inventory buffers.

Work the arithmetic on a conservative example. Suppose FlowForge's heat-treat line processes 1,800 parts per day at the constraint while downstream CMM inspection can clear 2,200 units per day. Increasing heat-treat throughput 8% without adding inspection capacity may only relocate the bottleneck and inflate WIP. Multiply queue delay by average margin per part to communicate dollar risk to executives who do not live in Gantt charts. Pair point estimates with guardrails: scrap rate, overtime hours, and customer premium freight.

Stakeholder conflict is normal. Greg Santos may push overtime to clear a automotive backlog while Nina Kowalski holds spending until lean kaizen (continuous small improvements, Japanese for "change for the better") tests finish. The CFO may push inventory cuts that lengthen setup-heavy campaigns. Capacity and Utilization gives you language to negotiate those tensions with capacity, quality, and forecast evidence rather than charisma.

Translate lessons to your own context by replacing FlowForge names while keeping structure. Pick one operations decision you face this quarter. Write the process boundary, constraint assumption, primary metric, guardrails, and kill criteria before changing the schedule. If you cannot write those elements, you are not ready to approve overtime or capital regardless of how urgent the email thread feels.

Technical mechanics and checks (worked patterns)

For capacity and utilization, FlowForge analysts show work the way finance shows reconciliations. A process capacity table lists resource, time per unit, units per hour, daily capacity at stated shift pattern, and a check that the bottleneck matches the lowest capacity step. A Little's Law table prints average WIP, throughput, and implied flow time with a check that $I = R \times T$ reconciles within rounding. A control-chart appendix lists subgroup size, center line, control limits, and rule violations before a line stop is authorized. A forecast table shows actual, forecast, absolute error, and cumulative bias by family.

Use plain-language statements before formulas. Example for capacity: process capacity equals the minimum capacity across serial steps unless parallel paths merge. FlowForge forbids ambiguous one-word metrics like efficiency without stating whether it means OEE, labor efficiency, or first-pass yield. Each definition implies different data collection and different managerial meaning.

For spreadsheet or ERP replication, write the grain first. Order-line tables suit on-time delivery. Operation-sequence tables suit routing-based capacity. Shift-level tables suit OEE losses. SKU-family tables suit forecast accuracy. FlowForge Components ties every lesson metric to a named owner on the operations review slide.

Common executive questions (and disciplined answers)

Executives ask short questions that require long disciplined answers. "Are we capacity constrained?" maps to bottleneck utilization, WIP shape, and overtime trend, not gut feel from the parking lot. "Is quality getting better?" maps to defect Pareto, SPC signals, and cost of poor quality, not one good week after a customer audit. "Can we trust the forecast?" maps to bias, MAPE (mean absolute percentage error), and forecast value added versus a naive baseline. "Why not just add a shift?" maps to demand permanence, training cost, and whether the constraint moves.

FlowForge's credible answer format for capacity and utilization is three bullets: recommendation, evidence strength (descriptive, tested, scaled), and next study if limitations matter. A fourth bullet lists what would falsify the recommendation within sixty days. That discipline prevents the operations team from becoming either a bottleneck or a rubber stamp.

Linking Capacity and Utilization to prior and next lessons in OPS 201

Operations fluency is cumulative. Capacity and Utilization in Unit 2 connects backward to definitions and forward to integrative decisions. When you read FlowForge examples, mark which numbers are structural (routing standards, shift calendars, contractual service levels) versus policy (safety stock targets, overtime triggers, inspection sampling rates). Mixing the two produces recommendations that work once and fail next quarter.

Nina Kowalski's team keeps a single-page operating system for each plant: strategic priorities from Unit 1, process facts from Unit 2, service and queue policies where customers wait, quality and lean cadence from Unit 4, planning horizons from Unit 5, and capital or outsourcing choices from Unit 6. Capacity and Utilization should slot into that page with an owner and review frequency. If it does not slot anywhere, it is trivia.

Practice teaching capacity and utilization aloud using only FlowForge nouns and one table. If your explanation requires generic "a factory," you have not yet transferred the lesson. Retry with 1,800 parts per day, 78% OEE, and a named OEM program deadline.

Applying Capacity and Utilization at FlowForge scale

When FlowForge Components evaluates capacity and utilization, VP Operations Nina Kowalski and Plant Manager Greg Santos start from operational facts: $215M revenue, 78% OEE (overall equipment effectiveness), 1.2% external defect rate, and 94% on-time delivery to OEM customers. The process analysis, flow metrics, and capacity review cadence is weekly on the Toledo shop floor and monthly with the CEO and CFO. A lesson concept that sounds abstract becomes concrete when tied to CNC cycle times, heat-treat queue lengths, and PPAP (production part approval process, the automotive quality gate before volume shipment) holds.

Consider how a one-point OEE improvement affects FlowForge. At 42 machining centers running three shifts, a single point of OEE often frees roughly $1.8M to $2.4M of effective capacity annually without new capital, depending on bottleneck mix and scrap rework rates. That is why capacity and utilization is not academic for Nina Kowalski; it is how the company funds automation without missing aerospace delivery windows.

The process analysis, flow metrics, and capacity workflow at FlowForge deliberately separates descriptive dashboards from causal improvement tests. A spike in WIP (work in process, partially completed units between operations) triggers a value-stream walk before overtime is approved. A quality escape triggers containment, root-cause analysis, and SPC (statistical process control, using control charts to distinguish common-cause from special-cause variation) review on the affected line. Forecast errors trigger aggregate-planning revisions before raw bar stock is purchased. Label outputs before they reach the executive committee: observation, tested mechanism, or scaled policy.

Document definitions alongside every operations metric tile. FlowForge's OEE formula specifies availability losses (planned maintenance versus breakdown), performance losses (speed versus standard cycle), and quality losses (scrap and rework at the constraint). On-time delivery excludes customer-approved pull-ins but includes contractual grace days. Defect rate is measured at OEM incoming inspection per million opportunities. When definitions live in a shared dictionary, the company builds institutional memory instead of re-debating the same spreadsheet every quarter.

Extended FlowForge scenario: cross-functional read

Imagine FlowForge's Q3 review for capacity and utilization. Finance asks whether a capacity investment clears hurdle rate given 8.2 inventory turns and rising interest expense. Commercial asks whether on-time delivery can hold at 94% if automotive mix shifts toward shorter lead-time programs. Quality asks whether the 1.2% external defect rate threatens PPAP status on a new aerospace cell. A weak process analysis, flow metrics, and capacity answer addresses only one function. A strong answer shows how evidence flows: process maps localize WIP buildup at heat treat, capacity models quantify constraint hours, control charts separate noise from special cause, and forecast error bands drive staffing and inventory buffers.

Work the arithmetic on a conservative example. Suppose FlowForge's heat-treat line processes 1,800 parts per day at the constraint while downstream CMM inspection can clear 2,200 units per day. Increasing heat-treat throughput 8% without adding inspection capacity may only relocate the bottleneck and inflate WIP. Multiply queue delay by average margin per part to communicate dollar risk to executives who do not live in Gantt charts. Pair point estimates with guardrails: scrap rate, overtime hours, and customer premium freight.

Stakeholder conflict is normal. Greg Santos may push overtime to clear a automotive backlog while Nina Kowalski holds spending until lean kaizen (continuous small improvements, Japanese for "change for the better") tests finish. The CFO may push inventory cuts that lengthen setup-heavy campaigns. Capacity and Utilization gives you language to negotiate those tensions with capacity, quality, and forecast evidence rather than charisma.

Translate lessons to your own context by replacing FlowForge names while keeping structure. Pick one operations decision you face this quarter. Write the process boundary, constraint assumption, primary metric, guardrails, and kill criteria before changing the schedule. If you cannot write those elements, you are not ready to approve overtime or capital regardless of how urgent the email thread feels.