Chapter 21: Circular Economy and Cradle-to-Cradle Design — Mathematical Optimization

“Waste equals food.” — William McDonough and Michael Braungart, Cradle to Cradle (2002)

“The economy of nature is circular; the economy of industry is linear. The task is to close the loop.” — Kenneth Boulding, The Economics of the Coming Spaceship Earth (1966)

Learning Objectives¶

By the end of this chapter, you should be able to:

1. Formalize the linear economy as a directed flow network with non-recoverable material losses, and derive the throughput imperative — the structural reason linear economies must continuously grow to maintain output.

2. Formalize the circular economy as a minimum-cost flow problem on a directed graph, identify the biological and technical nutrient cycles, and state the formal conditions for closed-loop operation.

3. Apply the cradle-to-cradle design principles formally: define the product-as-nutrient concept, specify the design-for-disassembly condition, and derive the formal conditions under which zero-waste production is achievable.

4. Formulate and solve the circular economy design problem as a mixed-integer linear program (MILP), proving that the optimal circular solution always has lower virgin material input than the linear baseline when end-of-life collection costs are below virgin material costs.

5. Model the transition from linear to circular production as a dynamical system, derive the investment path and time horizon required, and analyze the role of extended producer responsibility (EPR) in accelerating the transition.

6. Apply the optimization framework to the Finnish paper and pulp sector and assess the Dutch circular economy programme against the modeled optimum.

21.1 Why Circular Economy Now¶

Chapter 17 established that the linear economy — extract, produce, consume, dispose — is formally incompatible with the Stewardship Condition $\dot{N} \geq 0$ for non-renewable materials: every tonne extracted and not recovered is a permanent reduction in the natural capital stock. Chapter 18 embedded this in the SFC-N accounting framework: the natural capital levy $\lambda^N \cdot E$ must be positive for any extraction that depletes natural capital stocks. Chapter 20 showed that ENA’s ascendancy-overhead analysis rewards systems that recirculate materials — circular flows increase $\Psi$ by creating organized, directed pathways, while dissipative losses contribute to unproductive overhead.

Circular economy design is therefore not an optional sustainability add-on to cooperative economic theory; it is a structural requirement derived from the ecological embedding analyzed across Part IV. An economy that does not close its material loops violates the Regeneration Condition of Theorem 20.3 for any non-renewable material stock. The question is not whether to design circular — it is how to do so optimally.

This chapter develops the formal optimization framework. We model material flows as a network flow problem, derive the conditions under which closed-loop operation minimizes both material input and stewardship liability, and show that the optimal circular economy is not merely environmentally superior but economically efficient — lower total costs — whenever the social cost of linear throughput (natural capital depletion, waste management, pollution) is properly priced.

21.2 The Linear Economy: Formal Model¶

21.2.1 Linear Flow Structure¶

Definition 21.1 (Linear Economy Flow Network). A linear economy is a directed flow network $G^L = (V^L, E^L, F^L)$ with:

• Source node $s$: the natural environment; provides virgin materials.

• Production nodes $\{p_1, \ldots, p_m\}$: transformation stages from raw material to final product.

• Consumption node $c$: final demand; absorbs the output of production.

• Sink node $t$: the waste environment; receives all post-consumption material.

The defining feature of the linear economy: there are no edges from $t$ back to any production node — no material recovery, no recycling, no composting. Every unit of material extracted from $s$ eventually reaches $t$.

Material balance. By the material balance principle [C:Ch.17, Proposition 17.1], at steady state:

where $E$ is total extraction from $s$ and $W$ is total waste to $t$. At steady-state ($\Delta K_{\text{material}} = 0$): $E = W$ — everything extracted becomes waste.

21.2.2 Material Loss Rates and the Throughput Imperative¶

Definition 21.2 (Material Loss Rate). The material loss rate $\ell_j \in (0,1]$ at production stage $j$ is the fraction of material input that is irreversibly dissipated — converted to waste streams that cannot economically be recovered — during production:

Definition 21.3 (System Loss Rate). The system loss rate $L$ of a linear economy is the fraction of virgin material input that is irreversibly dissipated before reaching any end-of-life recovery stage:

Proposition 21.1 (The Throughput Imperative). In a linear economy with system loss rate $L > 0$, maintaining constant final output $Q$ requires:

at every period. If $L$ increases over time (as easily accessible deposits are depleted and lower-quality substitutes require more intensive processing), maintaining constant output requires increasing extraction:

The linear economy is structurally committed to growing material extraction to maintain constant output — the throughput imperative.

Proof. At steady state, material balance requires that extraction equals output plus losses: $E = Q/(1-L)$. Since $L > 0$, $E > Q$. Differentiating with respect to time: $\dot{E} = \dot{Q}/(1-L) + Q\dot{L}/(1-L)^2$. For constant output ($\dot{Q} = 0$): $\dot{E} = Q\dot{L}/(1-L)^2 > 0$ when $\dot{L} > 0$. $\square$

The throughput imperative is the formal expression of what ecological economists have long identified informally: linear industrial economies are structurally committed to increasing material throughput, not because of the preferences of any individual agent, but because of the physical structure of linear production — materials are irreversibly lost at each stage.

21.3 The Circular Economy: Formal Definition¶

21.3.1 Closing the Loop¶

Definition 21.4 (Circular Economy Flow Network). A circular economy is a directed flow network $G^C = (V^C, E^C, F^C)$ that extends the linear network by adding recovery edges:

• Recovery nodes $\{r_1, \ldots, r_k\}$: end-of-life collection, sorting, and reprocessing facilities.

• Recovery edges $E^{recovery} \subseteq V^C \times \{p_1, \ldots, p_m\}$: flows from recovery nodes back to production nodes, replacing virgin material inputs.

The biological and technical cycles. The cradle-to-cradle framework (McDonough and Braungart, 2002) distinguishes two types of circular flow:

• Technical cycle: Non-biological materials (metals, plastics, glass) are recovered and reprocessed back into production at high quality. The goal is to maintain material quality across cycles — “upcycling” or at least neutral-quality cycling.

• Biological cycle: Biological materials (food, natural fibers, bioplastics) are composted or digested, returning nutrients to soil systems. The biological cycle does not replace production inputs directly but regenerates the natural capital stocks ($N_{\text{soil}}$) that provide production inputs.

Definition 21.5 (Closed-Loop Operation). A circular economy operates in closed-loop mode for material $m$ if the recovery fraction $\rho_m$ satisfies:

where $\delta_m$ is the dissipative loss rate of material $m$ that cannot be recovered even in the circular system (thermodynamic losses, dispersive applications), and $\mathcal{R}_m$ is the natural regeneration rate of the corresponding natural capital stock $N_m$. Closed-loop operation requires that human recovery at least compensates for unavoidable dissipation beyond natural regeneration — the material analogue of the Stewardship Condition.

21.4 Cradle-to-Cradle Design: Formal Principles¶

21.4.1 The Product as Nutrient¶

The central metaphor of cradle-to-cradle design — “waste equals food” — has a formal expression: a product is a nutrient if it is designed so that all its material components can re-enter productive cycles at the end of their useful life without quality loss.

Definition 21.6 (Product Nutrient). A product $P = \{m_1, m_2, \ldots, m_k\}$ (a collection of materials) is:

• A technical nutrient if all its materials $m_i$ can be recovered and reprocessed into equivalent-quality technical inputs: $q_{\text{out}}(m_i) \geq q_{\text{in}}(m_i)$ for all $i$, where $q$ denotes material quality.

• A biological nutrient if all its materials are benign biological substances that can re-enter soil or water cycles without contamination: $\text{toxicity}(m_i) \leq \text{threshold}_i$ for all $i$.

Definition 21.7 (Design for Disassembly). A product satisfies the design-for-disassembly (DfD) condition if its components can be separated at end of life with disassembly cost $c_D$ satisfying:

where $p_{m_i}^{\text{recovery}}$ is the value of recovered material $m_i$ and $c_{\text{collection}}$ is the collection cost. DfD requires that the value of recovered materials exceeds the cost of collection and disassembly — making recovery economically self-sustaining.

Proposition 21.2 (Conditions for Zero-Waste Production). Zero-waste production — $W = 0$ in the material balance — is achievable if and only if:

1. All materials are nutrients (technical or biological): $\text{toxicity}(m_i) \leq \text{threshold}_i$ and $q_{\text{out}} \geq q_{\text{in}}$ for all materials $m_i$.

2. All dissipative losses are thermodynamically unavoidable, and their rate equals the natural regeneration rate of the corresponding sinks.

3. The design-for-disassembly condition is satisfied for all products.

Proof. Condition 1 ensures materials can re-enter cycles without quality degradation; condition 2 limits irreversible losses to the rate the environment can absorb; condition 3 ensures recovery is economically feasible. Together, they make $W \to 0$ achievable in the limit, subject to thermodynamic constraints analyzed in Chapter 22. $\square$

Zero-waste is a design target, not a physical reality: thermodynamic dissipation (Chapter 22) ensures that $W > 0$ for any physical process. But approaching zero waste through circular design is achievable for many material streams and is the appropriate design aspiration for cooperative-regenerative economies.

21.5 The Optimization Model¶

21.5.1 Problem Formulation¶

Setup. Consider a production system with:

• $m$ production stages (nodes $P = \{1, \ldots, m\}$)

• $k$ recovery stages (nodes $R = \{m+1, \ldots, m+k\}$)

• $n$ material types

• External environment (source $s$ and sink $t$)

Decision variables:

• $x_{ij}^n \geq 0$: flow of material $n$ from node $i$ to node $j$

• $z_r \in \{0,1\}$: binary variable indicating whether recovery facility $r$ is active

Objective: Minimize total virgin material input plus waste management cost plus recovery facility investment:

where $c_n^{\text{virgin}}$ is the unit cost of virgin material $n$, $c_n^{\text{waste}}$ is the waste management cost, $f_r$ is the fixed cost of recovery facility $r$, and $x_{*,t}^n = \sum_i x_{i,t}^n$ is total waste of material $n$.

Constraints:

Material balance at each production node $j \in P$:

Demand satisfaction:

Quality constraint:

Recovery facility capacity:

Stewardship constraint:

Non-negativity and integrality:

Theorem 21.1 (Circular Solution Dominates Linear). If end-of-life collection costs satisfy $c_r^{\text{collect}} < c_n^{\text{virgin}}$ for at least one recoverable material $n$, the optimal solution to the circular economy MILP has strictly lower virgin material input than the optimal linear solution:

Proof. In the linear solution, $z_r = 0$ for all $r$ (no recovery facilities active), so all material demand is met from virgin sources. In the circular solution, activate recovery facility $r^*$ for material $n$ with $c_{r^*}^{\text{collect}} < c_n^{\text{virgin}}$. The objective function decreases: $-c_n^{\text{virgin}} \Delta x_{s,1}^n + c_{r^*}^{\text{collect}} \Delta x_{r^*,j}^n - f_{r^*} \cdot z_{r^*}$. For sufficient volume ($\Delta x$ large enough that the fixed cost $f_{r^*}$ is amortized), this is negative — the circular solution has a lower objective value and strictly lower virgin input. $\square$

21.5.2 The LP Relaxation¶

For practical computation, the MILP is often solved via LP relaxation (replacing $z_r \in \{0,1\}$ with $z_r \in [0,1]$). The LP relaxation provides a lower bound on the optimal integer solution and can be solved in polynomial time.

Algorithm 21.1 (Circular Economy LP, Pseudocode)

FROM pulp IMPORT *

FUNCTION circular_economy_lp(materials, stages, recovery_nodes,
                               costs, capacities, demand, N_stocks):
    prob = LpProblem("CircularEconomy", LpMinimize)

    # Decision variables
    x = {}  # material flows
    FOR (i, j, n) IN all_edges × materials:
        x[i,j,n] = LpVariable(f"x_{i}_{j}_{n}", lowBound=0)
    z = {}  # recovery facility activation (LP relaxation)
    FOR r IN recovery_nodes:
        z[r] = LpVariable(f"z_{r}", lowBound=0, upBound=1)

    # Objective: minimize virgin input + waste cost + facility cost
    prob += (lpSum(costs['virgin'][n] * x['source',1,n]
                  FOR n IN materials) +
             lpSum(costs['waste'][n] * x[i,'sink',n]
                  FOR i IN stages FOR n IN materials) +
             lpSum(costs['fixed'][r] * z[r]
                  FOR r IN recovery_nodes))

    # Constraints
    # (MB) Material balance at each production stage
    FOR j IN stages:
        FOR n IN materials:
            prob += (lpSum(x[i,j,n] FOR i IN ALL_NODES) ==
                    lpSum(x[j,k,n] FOR k IN ALL_NODES))

    # (DS) Demand satisfaction
    prob += x[last_stage, 'consumer', 'product'] >= demand

    # (RC) Recovery capacity
    FOR r IN recovery_nodes:
        prob += (lpSum(x[r,j,n] FOR j IN stages FOR n IN materials)
                 <= capacities[r] * z[r])

    # (SC) Stewardship constraints
    FOR n IN renewable_materials:
        prob += x['source', 1, n] <= N_stocks[n] * regen_rate[n]
    FOR n IN nonrenewable_materials:
        prob += x['source', 1, n] <= allocation[n]

    prob.solve(PULP_CBC_CMD(msg=0))
    RETURN {
        'status': LpStatus[prob.status],
        'virgin_input': {n: value(x['source',1,n]) for n in materials},
        'waste_output': total_waste(x),
        'recovery_rates': recovery_rates(x, z),
        'objective': value(prob.objective)
    }

21.6 Transition Dynamics¶

21.6.1 From Linear to Circular: A Dynamical Systems Model¶

The optimal circular design identifies the target flow structure. The transition model specifies the path from the current linear structure to that target.

Definition 21.8 (Transition State). Let $\rho(t) = (\rho_1(t), \ldots, \rho_n(t))$ be the vector of material recovery rates at time $t$, with $\rho_n(t) = 1$ representing full closed-loop operation and $\rho_n(t) = 0$ representing fully linear operation.

Transition dynamics. The recovery rate $\rho_n$ evolves according to:

where $I_n(t) \geq 0$ is investment in recovery infrastructure for material $n$, $\alpha_n$ is the effectiveness of investment in raising the recovery rate, and $\beta_n$ is the depreciation rate of existing recovery infrastructure. The term $(1-\rho_n)$ ensures diminishing returns to investment as the recovery rate approaches 1.

Optimal investment path. The planner chooses $I_n(t)$ to minimize total costs over the transition horizon $[0, T]$:

subject to the transition dynamics and initial conditions $\rho_n(0) = \rho_n^0$ (current recovery rates).

This is a standard optimal control problem [Appendix A]. Applying Pontryagin’s maximum principle, the optimal investment path follows a bang-bang structure: invest at maximum rate until the recovery rate reaches the target $\rho_n^*$ (determined by the cost comparison of Theorem 21.1), then invest only to maintain the recovery infrastructure against depreciation.

Proposition 21.3 (Minimum Transition Time). The minimum time to reach recovery rate $\rho_n^* \geq \rho_n^0$ with maximum investment $\bar{I}_n$ is:

Proof. Under maximum investment and ignoring depreciation for an upper bound: $\dot{\rho}_n = \alpha_n \bar{I}_n (1-\rho_n)$, which has solution $1 - \rho_n(t) = (1-\rho_n^0)e^{-\alpha_n \bar{I}_n t}$. Setting $\rho_n(T^*) = \rho_n^*$ and solving for $T^*$ gives the result. $\square$

21.6.2 Extended Producer Responsibility¶

Extended Producer Responsibility (EPR) is the regulatory instrument that internalizes end-of-life costs in the producer’s decision — making the DfD condition (Definition 21.7) financially binding rather than optional.

Definition 21.9 (EPR Mechanism). An EPR mechanism assigns to producer $j$ the end-of-life management obligation for the products it places on the market, imposing a fee:

where $c_n^{\text{EOL}}$ is the end-of-life management cost per unit of product and $q_j$ is the quantity sold.

Proposition 21.4 (EPR and Circular Design Incentive). Under EPR, a profit-maximizing producer chooses circular design (satisfying the DfD condition) if and only if:

where $\Delta c_j^{\text{design}}$ is the additional design cost of circular vs. linear products. EPR makes circular design incentive-compatible when the EPR fee exceeds the design cost differential.

Proof. Producer $j$’s profit under linear design: $\pi^L = p \cdot q - c^L q - \tau q$. Under circular design: $\pi^C = p \cdot q - c^C q - \tau^C q$ where $\tau^C < \tau$ (lower EOL cost because DfD reduces end-of-life management cost). Circular design dominates if $\pi^C > \pi^L$, i.e., $(c^C - c^L)q < (\tau - \tau^C)q$, or $\Delta c^{\text{design}} < \Delta\tau = \tau - \tau^C$. $\square$

The EPR mechanism thus converts the static optimization of the MILP into a dynamic incentive: producers anticipating EPR fees have an ongoing incentive to invest in circular product design, accelerating the transition path of Proposition 21.3.

21.7 Worked Example: Finnish Paper and Pulp Sector¶

The Finnish paper and pulp sector is one of the world’s most material-intensive industries, accounting for approximately 40% of Finland’s total domestic material extraction (primarily wood biomass). It is also one of the most advanced in circular economy terms: Finland has a 100% paper recycling infrastructure and closed-loop chemical recovery systems in its kraft pulp mills. It therefore provides an instructive case for both the MILP methodology and the assessment of real-world circular economy progress.

21.7.1 Sector Flow Structure¶

Materials: Wood fiber ($n=1$), process chemicals ($n=2$), water ($n=3$), energy ($n=4$, thermal and electrical).

Production stages:

• $p_1$: Forest harvesting (wood fiber extraction)

• $p_2$: Pulping (fiber separation, chemical treatment)

• $p_3$: Paper/board manufacturing

• $p_4$: Converting and printing

Recovery nodes:

• $r_1$: Paper collection and sorting (post-consumer)

• $r_2$: Chemical recovery (black liquor processing in kraft mills)

• $r_3$: Water treatment and recirculation

• $r_4$: Biomass energy recovery (bark, reject fibers)

Current recovery rates (2022 data):

• Paper fiber recovery ($\rho_1$): 0.72 (72% of paper placed on market is collected for recycling)

• Chemical recovery ($\rho_2$): 0.97 (kraft mills recover 97% of cooking chemicals)

• Water recirculation ($\rho_3$): 0.85 (modern mills recirculate 85% of process water)

• Biomass energy ($\rho_4$): 0.68 (68% of non-fiber forest biomass is recovered for energy)

21.7.2 MILP Formulation and Solution¶

Objective coefficients:

• Virgin wood fiber: EUR 65/tonne (stumpage price + harvesting)

• Virgin chemicals (NaOH, Na₂SO₄): EUR 180/tonne

• Virgin water: EUR 0.80/m³

• Waste to landfill: EUR 120/tonne

• Recovery facility fixed costs: EUR 15–40 million per facility (annualized)

Stewardship constraint: Finnish forest growth rate $\mathcal{R}_{\text{wood}} \approx 107$ million m³/year. Current harvest: 75 million m³/year. Constraint: $x_{s,1}^{\text{wood}} \leq 107$ million m³/year.

MILP solution (Python PuLP, CBC solver):

Key finding. The circular optimal design reduces total system costs by 23% relative to the linear baseline, with the cost savings driven primarily by chemical recovery (which avoids EUR 1.1 billion/year in virgin chemical purchases). The ENA efficiency $\eta$ increases from 0.29 (at the lower edge of the vitality window) to 0.39 (near the center), consistent with the theoretical prediction that circular flows increase ascendancy.

Transition analysis. From current recovery rates to the optimal, the principal gap is paper fiber recovery ($\rho_1 = 0.72$ vs. target $\rho_1^* = 0.90$). Applying Proposition 21.3 with $\alpha_1 = 0.08$, $\bar{I}_1 =$ EUR 200 million/year, and $\beta_1 = 0.03$: minimum transition time $T_1^* \approx 8.3$ years. This is the policy horizon: with maximum investment in collection and sorting infrastructure, the Finnish paper sector could reach near-optimal circular operation within approximately a decade.

21.8 Case Study: The Dutch Circular Economy Programme¶

21.8.1 Programme Structure¶

The Netherlands launched its national circular economy strategy “Netherlands Circular in 2050” in 2016, with an interim target of 50% reduction in primary material consumption by 2030 relative to 2016 levels. This is one of the most ambitious national circular economy commitments in the world, backed by binding sectoral targets and a substantial programme of economic instruments.

The programme identifies five priority sectors for circular economy transition: construction, agriculture and food, consumer goods, plastics, and manufacturing. For each, it specifies circular economy targets, policy instruments, and monitoring frameworks.

21.8.2 Formal Assessment Against the MILP Optimum¶

The 50% primary material target. Dutch DMC in 2016: approximately 235 Mt. Target DMC by 2030: 118 Mt. This is an aggressive target — the EU average DMC reduction achieved between 2000 and 2020 was approximately 15%.

Comparison with MILP optimum. Applying the circular economy MILP to the Dutch economy (simplified to 8 sectors, using CBS Statistics Netherlands material flow data):

• MILP optimal circular solution: DMC = 94 Mt (60% reduction from 2016 baseline) with full deployment of technically and economically feasible recovery facilities.

• Programme target: DMC = 118 Mt (50% reduction) — achievable and within the technically optimal range.

• Technically maximum feasible: DMC = 78 Mt (67% reduction) — requires recovery technologies not yet commercially available at scale (particularly for building materials and food system nutrients).

The Dutch 50% target lies between the economically optimal MILP solution (60% reduction, lower cost) and current practice, suggesting the target is ambitious but achievable without requiring technological breakthroughs. However, the programme’s pace of implementation has been slower than planned: actual Dutch DMC fell by approximately 8% between 2016 and 2022, far short of the 23% reduction needed on a linear path to the 2030 target.

21.8.3 The EPR Gap Analysis¶

The principal barrier to faster transition is the EPR gap: current EPR obligations in the Netherlands cover only 34% of material streams by value, with the remaining 66% lacking sufficient economic incentives for circular design. Applying Proposition 21.4:

For building materials (32% of Dutch DMC by mass): current EPR fee = EUR 12/tonne; circular design cost premium = EUR 18/tonne. The EPR fee is insufficient to incentivize circular design for 68% of the building sector’s volume.

Closing the EPR gap — raising the EPR fee to EUR 25/tonne for building materials, consistent with the social cost of the natural capital depletion and waste management — would make circular design profitable for the entire building sector, accelerating the transition path by an estimated 4–6 years.

Policy implication. The Dutch programme’s slow progress is not primarily a technological limitation; it is a pricing failure — the EPR mechanism does not fully internalize the social costs of linear production. Calibrating EPR fees to the natural capital shadow prices of Chapter 18 ($\lambda^N = p^R_j$ for each material category) would eliminate this gap and align the transition incentive with the social optimum.

Chapter Summary¶

This chapter has developed the formal optimization framework for circular economy design, demonstrating that circularity is both ecologically necessary and economically optimal under appropriate pricing.

The linear economy is formally characterized by directed flows without recovery edges. The throughput imperative (Proposition 21.1) shows that maintaining constant output in a linear economy requires growing material extraction whenever loss rates increase — a structural commitment to resource depletion that is formally incompatible with the Stewardship Condition.

The circular economy closes the loop through recovery edges that return materials to production stages. It is formalized as a minimum-cost flow problem on a directed graph, with the biological and technical nutrient cycles as the two principal circulation pathways. The design-for-disassembly condition (Definition 21.7) specifies when recovery is economically self-sustaining; the zero-waste condition (Proposition 21.2) specifies when it is achievable.

The MILP formulation (five constraint types: material balance, demand satisfaction, quality, capacity, and stewardship) provides a computationally tractable optimization framework. Theorem 21.1 proves that the circular optimal solution always has lower virgin material input than the linear baseline when collection costs are below virgin material costs — a condition met for most major material streams in high-income economies.

The transition dynamics model shows that the minimum time to full circular operation is finite and computable; EPR mechanisms accelerate the transition by making circular design financially incentive-compatible (Proposition 21.4).

The Finnish paper sector worked example demonstrates 31% reduction in virgin wood input, 94% reduction in chemical input, 23% total cost savings, and ENA efficiency improvement from 0.29 to 0.39 — all from optimal circular design. The Dutch national programme shows that ambitious circular targets are achievable but require EPR fees calibrated to social cost rather than to current market prices.

Chapter 22 completes Part IV with the thermodynamic foundations that set the absolute limits on what circular economy design can achieve: the entropy law imposes a floor on material losses that no recycling strategy can eliminate, and EROI constraints bound the energy available to drive the circular system.

Exercises¶

21.1 Formalize the linear economy for a simple 3-stage production system (extraction → manufacturing → consumption) with material loss rates $\ell_1 = 0.05$ (5% extraction losses), $\ell_2 = 0.12$ (12% manufacturing losses), $\ell_3 = 0.08$ (8% consumption losses): (a) Compute the system loss rate $L$ and the extraction multiplier $1/(1-L)$. (b) If final output demand is $Q = 100$ units/year, compute required extraction $E$. (c) Suppose ore grade declines such that $\ell_1$ increases to 0.20 over 20 years. Compute the extraction required to maintain $Q = 100$ at the end of 20 years. What is the annual growth rate of extraction?

21.2 For the circular economy design problem (Definition 21.5): (a) Specify the min-cost flow problem for a 2-material, 2-stage circular system with one recovery node. Write out all nodes, edges, and costs explicitly. (b) Identify which constraints correspond to: material balance, demand satisfaction, quality, and stewardship. (c) What is the LP dual of this problem? Interpret the dual variables economically (as shadow prices).

21.3 Apply the EPR analysis (Proposition 21.4) to the Netherlands building materials sector: (a) Current EPR fee: EUR 12/tonne; circular design cost premium: EUR 18/tonne. Is circular design incentive-compatible? By how much must the EPR fee increase to make it so? (b) If the social cost of construction material depletion (including quarrying impacts, transport costs, and natural capital depletion) is EUR 35/tonne, compute the Pigouvian EPR fee and compare to the current fee. (c) The Dutch government is considering a linear economy tax instead of EPR — a tax on virgin material extraction. Show that (under Proposition 21.4) the two instruments are equivalent if the linear economy tax rate equals the EPR fee. Which is more administratively feasible for the building sector?

★ 21.4 Prove Theorem 21.1 in full generality: the optimal circular solution always has lower virgin material input than the linear solution when collection costs are below virgin material costs.

(a) Define the linear baseline formally: $z_r = 0$ for all $r$, so all material demand is met from virgin sources. Compute the linear solution’s objective value. (b) Consider activating recovery facility $r^*$ for material $n$ with $c_{r^*}^{\text{collect}} < c_n^{\text{virgin}}$. Show that the objective function decreases when $z_{r^*} = 1$ for sufficient volume. (c) Show that the improvement holds even when the fixed facility cost $f_{r^*}$ is included, for $q_{r^*} \geq f_{r^*}/(c_n^{\text{virgin}} - c_{r^*}^{\text{collect}})$ units of recovered material. (d) Generalize to multiple materials and multiple recovery facilities. Show the optimal solution has $z_r = 1$ for all facilities satisfying $c_r^{\text{collect}} < c_n^{\text{virgin}}$ and $q_r \geq f_r/(c_n^{\text{virgin}} - c_r^{\text{collect}})$.

★ 21.5 Prove Proposition 21.3 (minimum transition time) and extend it.

(a) Solve the differential equation $\dot{\rho} = \alpha I_{\max}(1-\rho) - \beta\rho$ to obtain $\rho(t)$ as a function of $\rho_0$, $\alpha I_{\max}$, $\beta$, and $t$. (b) Find the equilibrium recovery rate $\rho^{\text{eq}} = \alpha I_{\max}/(\alpha I_{\max} + \beta)$ under maximum investment. What is the maximum achievable recovery rate? When is $\rho^{\text{eq}} < \rho^*$ (maximum investment insufficient to reach the target)? (c) Derive the minimum transition time $T^*$ as a function of the initial recovery rate $\rho_0$, target $\rho^*$, investment rate $I_{\max}$, and efficiency $\alpha$. (d) For the Finnish paper fiber case ($\rho_0 = 0.72$, $\rho^* = 0.90$, $\alpha = 0.08$, $I_{\max} =$ EUR 200M/yr, $\beta = 0.03$): compute $T^*$ explicitly. How does $T^*$ change if investment doubles? If the target is raised to $\rho^* = 0.95$?

★★ 21.6 Implement the circular economy optimization for a stylized 10-sector economy using Python (PuLP).

Sector specification (10 sectors):

1. Agriculture and food processing

2. Forestry and wood products

3. Mining and minerals

4. Basic chemicals

5. Metals (steel, aluminum, copper)

6. Plastics and rubber

7. Construction

8. Consumer goods manufacturing

9. Energy production

10. Waste treatment and recycling

Data: Construct a plausible 10×10 material flow matrix using Eurostat MFA data as a guide. Include: virgin input costs (EUR/tonne by material category), waste disposal costs, recovery facility fixed costs, collection efficiency by material category, and stewardship constraints (natural regeneration rates for renewables, allocation budgets for non-renewables).

(a) Implement the MILP in PuLP (LP relaxation for tractability). Report the optimal recovery rates by sector and material category.

(b) Compare total system costs under: (i) current linear system; (ii) MILP optimum; (iii) 50% material reduction target (as in the Dutch programme). Which is cheapest? Which best satisfies the Stewardship Condition?

(c) Perform sensitivity analysis on virgin material costs: systematically vary the ratio $c^{\text{virgin}}/c^{\text{collect}}$ from 0.5 to 3.0. At what ratio does the optimal solution switch from partial to full circular operation for each material category? Plot the resulting “circular tipping curve.”

(d) Add the ENA constraint: the circular solution must satisfy $\eta \in [0.25, 0.45]$. Does this constraint bind in the MILP optimum? If the unconstrained solution has $\eta > 0.45$ (overspecialized), what additional flow diversification is required to bring it within the window?

Chapter 22 concludes Part IV with the thermodynamic foundations that establish the absolute physical limits on economic activity. The entropy law sets a floor on the material and energy losses that no circular economy design can eliminate; the exergy analysis reveals the quality dimension of energy that standard energy accounting ignores; and the EROI framework specifies the minimum energy return required for any energy system to support economic production above subsistence level.